Social Capital, Trust, and Firm Performance: The Value of Corporate Social Responsibility During The Financial Crisis

THE JOURNAL OF FINANCE • VOL. LXXII, NO.
4 • AUGUST 2017
Social Capital, Trust, and Firm Performance:

The Value of Corporate Social Responsibility
during the Financial Crisis
KARL V. LINS, HENRI SERVAES, and ANE TAMAYO∗
ABSTRACT
During the 2008–2009 financial crisis, firms with high social capital, as measured by
corporate social responsibility (CSR) intensity, had stock returns that were four to
seven percentage points higher than firms with low social capital. High-CSR firms
also experienced higher profitability, growth, and sales per employee relative to low-
CSR firms, and they raised more debt. This evidence suggests that the trust between
a firm and both its stakeholders and investors, built through investments in social
capital, pays off when the overall level of trust in corporations and markets suffers a
negative shock.
“The present financial crisis springs from a catastrophic collapse in confi-

dence . . . Financial markets hinge on trust, and that trust has eroded.”
—Joseph Stiglitz (2008)
“The fundamental problem isn’t lack of capital. It’s lack of trust. And
without trust, Wall Street might as well fold up its fancy tents.”
—Former U.S. Labor Secretary Robert Reich (2008)
∗ Karl V. Lins is at the University of Utah. Henri Servaes is at London Business School, CEPR,
and ECGI. Ane Tamayo is at the London School of Economics and Political Science (LSE). The
authors have no conflicts of interest to disclose. We would like to thank Taylor Begley, Colin Clubb,
Joao Cocco, Mike Cooper, James Dow, Alex Edmans, Christopher Hennessy, Ioannis Ioannou,
Ralph Koijen, Jean-Marie Meier, Yuval Millo, Michael Roberts (the Editor), Kelly Shue, Rui Silva,
Hannes Wagner, Yao Zeng, an anonymous Associate Editor, an anonymous referee, and seminar
participants at City University, Erasmus University, ESSEC, HEC Paris, INSEAD, King’s College,
London Business School, London School of Economics, Tilburg University, University of Bristol,
University of Edinburgh, University of Leicester, University of Melbourne, University of New South
Wales, University of Southampton, University of Sydney, WHU Otto Beisheim, the French Finance
Association, London Business School Summer Finance Symposium, the International Accounting
Research Symposium at the Fundación Ramón Areces, the International Corporate Governance
Conference at Hong Kong Baptist University, and the University of Cambridge Financial Account-
ing Symposium for helpful comments and discussions. We would also like to thank the ECGI for
the 2016 Standard Life Investments Finance Working Paper Prize. Dimas Fazio provided excellent
research assistance.
DOI: 10.1111/jofi.12505
1785
1786 The Journal of FinanceR
“The global financial and economic crisis has done a lot of harm to the
public trust in the institutions, the principles and the concept itself of the
market economy.”
—OECD Secretary General Angel Gurria (2009)
“Something important was destroyed in the last few months of 2008. It is

an asset crucial to production, even if it is not made of bricks and mortar . . .
This asset is trust.”
—Paolo Sapienza and Luigi Zingales (2012, p. 123)
The financial crisis highlighted the importance of trust for well-functioning
markets and financial stability, but discussions on the role of trust and, more
generally, social capital in economic life are not new. Already in 1972, Arrow
argued that “virtually every commercial transaction has within itself an ele-
ment of trust” (p. 357), and suggested that much of the economic backwardness
in the world might be due to the lack of mutual confidence. In line with this
view, Putnam (1993) shows that higher social capital societies, in which trust is
greater, display higher economic development (see also Fukuyama (1995), La
Porta et al. (1997), and Knack and Keefer (1997)). Focusing on capital markets,
Guiso, Sapienza, and Zingales (2004, 2008) document that trust derived from
greater social capital allows for more stock market participation. These studies
and other related work demonstrate the importance of social capital and trust
from a macroeconomic perspective. However, the extent to which social capital
and trust impact firm performance is relatively unexplored in the literature.
The objective of this paper is to address this question.
Empirical identification of the effect of trust and, more generally, social capi-
tal on firm performance is challenging. First, social capital is a broadly defined
concept, often encompassing trust and cooperative norms (e.g., Scrivens and
Smith (2013)), and hence its measurement is not straightforward. Second, with-
out exogenous variation in firm-level social capital, it is difficult to attribute
changes in performance to changes in social capital.
To address the first challenge, we focus on a firm’s Corporate Social Respon-
sibility (CSR) activities as a measure of its social capital, following recent work
in economics (Sacconi and Degli Antoni (2011)) suggesting that a firm’s CSR
activities are a good proxy for its social capital, and also the widespread view
among practitioners and corporations that a firm’s CSR activities generate
social capital and trust.1
To address the second challenge, we employ the 2008–2009 financial cri-
sis, a period during which public trust in corporations, capital markets, and
institutions declined unexpectedly.2 If a firm’s social capital helps build stake-
holder trust and cooperation (Putnam (1993)), it should pay off when being
1 Following the financial crisis, many corporations have emphasized the importance of a firm’s
social capital, driven by its CSR investments, in rebuilding stakeholder trust. However, the prac-
titioner view that CSR helps build trust predates the financial crisis (Fitzgerald (2003)).
2 The notion that the crisis led to a decline in public trust in corporations is corroborated by
surveys such as the 2009 Edelman Trust Barometer, which shows that 62% of respondents from a
Social Capital, Trust, and Firm Performance 1787
trustworthy is more valuable, such as in an unexpectedly low-trust period.

From a shareholder perspective, if high social capital firms are perceived as
more trustworthy, investors may place a valuation premium on these firms
when overall trust in companies is low (see Guiso, Sapienza, and Zingales
(2008)), as in the 2008–2009 financial crisis. From a stakeholder perspective,
the reciprocity concept often discussed in studies of social capital (i.e., the idea
that “I will be good to you because I believe you will be good to me at some
point in the future”) suggests that stakeholders (e.g., employees, customers,
suppliers, and the community at large) are more likely to help high social cap-
ital firms weather a negative shock, given that such firms displayed greater
attention to, and cooperation with, stakeholders in the past.
To test whether firm-level social capital pays off during a crisis of trust, we ex-
amine the performance of 1,673 nonfinancial firms with CSR data available on
the MSCI ESG Stats database (formerly known as KLD) over the August 2008
to March 2009 financial crisis period. In regressions that control for a wide vari-
ety of factors and firm characteristics (including governance and transparency),
we find that firms that entered the crisis period with high CSR ratings have
significantly higher (between four and seven percentage points) crisis-period
stock returns than those that entered it with low CSR ratings. The economic
importance of social capital in explaining stock returns is at least half as large
as the effect of cash holdings and leverage, financial variables previously shown
to affect crisis-period returns (Duchin, Ozbas, and Sensoy (2010) and Almeida
et al. (2012)). This result highlights the importance of expanding the focus
beyond financial capital when attempting to understand the determinants of
firm-level performance during a crisis of trust.
To alleviate concerns that the stock market outperformance we observe is
due to some factor other than a shock to trust, we conduct three further tests.
First, we investigate the association between CSR and stock returns during the
Enron/Worldcom crisis of the early 2000s, a period during which widespread
revelation of fraud undermined investor confidence in the U.S. stock market.
We find that high-CSR firms also earned excess returns relative to low-CSR
firms during this period. Second, we investigate whether our results are driven
by the decline in the supply of credit that firms faced during the financial crisis,
rather than by a decline in market-wide trust. Specifically, we test whether CSR
is related to stock returns in the period July 2007 through July 2008, when
there was a shock to the credit supply but no shock to the importance of trust.
We find no significant relation between CSR and stock returns during this
earlier period of the crisis. Third, we examine whether the relation between
CSR and crisis-period returns is stronger in high-trust regions, as identified in
the 2006 General Social Survey. We find that this is indeed the case.
It is possible, of course, that high-CSR firms also outperform low-CSR firms
during noncrisis periods (e.g., Edmans (2011)). To assess this possibility, we
examine whether the superior performance of high-CSR firms extends to
survey in 20 countries had lower trust in corporations in the aftermath of the financial crisis (for
respondents from the United States, this figure is 77%).
periods of economic growth or economic recovery using firm fixed effects models
that test the relation between CSR and firm performance before, during, and
after the crisis. These models show that CSR has a positive impact on returns
only during the crisis period, and that this effect is not due to time-invariant
unobservable firm characteristics.
We next seek to identify the mechanisms behind the outperformance of high-
CSR firms by examining firms’ profitability and productivity as well as their
capital raising during the crisis. We find that high-CSR firms have higher
profitability and gross margins, and experience higher sales growth, than other
firms during the crisis. They also have higher sales per employee and are able
to raise more debt. These results are consistent with stakeholder and investor
commitment to help firms deemed to be more trustworthy during the crisis.
Collectively, the findings that investors assign a premium to high-CSR firms
during a crisis of trust and that real effects take place at the firm level during
this time indicate that greater social capital maps into higher returns at the
microeconomic level. From a firm’s perspective, our results indicate that the
benefits that accrue to firms from building social capital through CSR activities
outweigh the costs of these activities when trust declines unexpectedly. As
such, investment in social capital can be thought of as an insurance policy
that pays off when investors and the economy at large face a severe crisis of
confidence and when the reward for being identifiably trustworthy increases
markedly. Our results thus highlight an enhanced insurance benefit of CSR
that goes beyond the notion that CSR acts as insurance against idiosyncratic
firm-specific legal risk (see, e.g., Godfrey, Merrill, and Hansen (2009), Minor
(2015), and Hong and Liskovich (2016)).
While our focus is on the impact of social capital on firm performance during
a shock to trust, our research design allows us to sidestep typical endogeneity
concerns that make it difficult to identify whether CSR activities impact firm
value, despite much research on this issue.3 In our natural experiment, the
exogenous financial shock disrupts the equilibrium, while levels of CSR remain
fixed, at least in the short term. This allows us to directly observe how investors
adjust their valuations of firms with differing attitudes toward CSR. Thus, this
paper also makes a contribution to the literature investigating whether CSR is
value-enhancing for shareholders. We recognize, however, that we do not have
exogenous variation in the levels of CSR, which limits the inferences we can
draw about the impact of CSR on performance during normal times.
The remainder of the paper is structured as follows. Section I discusses in
more detail the theoretical motivation behind our proxies and tests. Section II
discusses our data and summary statistics. In Section III, we analyze whether
CSR ratings impact stock returns during the crisis and conduct robustness
3 While much of the literature described thus far suggests that shareholders can derive value
from CSR investments, another strand of the literature argues that CSR investments could stem
from agency conflicts between managers and shareholders (see, e.g., Cheng, Hong, and Shue
(2016) for evidence that supports the agency view, and Ferrell, Liang, and Renneboog (2016) and
Albuquerque, Durnev, and Koskinen (2015) for evidence that does not). Margolis, Elfenbein, and
Walsh (2009) and Kitzmueller and Shimshack (2012) provide surveys of the CSR literature.
tests. In Section IV, we investigate several mechanisms that may explain the
excess performance of high-CSR firms. Section V concludes the paper.
I. Trust, Social Capital, and Corporate Social Responsibility

A. Trust and Social Capital
Over the last 20 years, the terms “social capital” and “trust” have become
increasingly popular in the economics and finance literature (Putnam (1993,
2000), Knack and Keefer (1997), La Porta et al. (1997), Guiso, Sapienza, and
Zingales (2004, 2008)).4 Often used indistinctly, both concepts are somewhat
abstract, although social capital is arguably the harder one to define due to its
multidimensional nature.
Trust is often understood as “the expectation that another person (or institu-
tion) will perform actions that are beneficial, or at least not detrimental, to us
regardless of our capacity to monitor those actions . . . so that we will consider
cooperating with him [the institution]” (Sapienza and Zingales (2012, p. 124),
based on Gambetta (1988)). This definition highlights the probabilistic nature
of trust (e.g., Gambetta (1988)), the concept of cooperation (e.g., Fukuyama
(1995) and La Porta et al. (1997)), and the inability to monitor others’ actions
ex ante (e.g., Dasgupta (1988)).5
Social capital is a broader concept. For example, Putnam (1993, 2000) views
social capital as “a propensity of people in a society to cooperate to produce
socially efficient outcomes” (La Porta et al. (1997, p. 333)) and highlights “the
norms of reciprocity and trustworthiness” that arise from connections among
individuals. A recent OECD paper (Scrivens and Smith (2013)) decomposes
social capital into four dimensions, with the intent of facilitating the devel-
opment of empirical measures: (i) personal relationships, (ii) social network
support, (iii) civic engagement, and (iv) trust and cooperative norms.6 The no-
tion of social capital that we explore, like much of the work in economics and
finance (e.g., Putnam (1993, 2000), Fukuyama (1995), Knack and Keefer (1997),
La Porta et al. (1997), Guiso, Sapienza, and Zingales (2004, 2008)), is mostly
related to the last two interpretations of the OECD.
The civic engagement aspect of social capital refers to the activities through
which agents contribute positively to the community and social life (e.g., vol-
unteering, political participation, donations; Guiso, Sapienza, and Zingales
(2011), Scrivens and Smith (2013)). Civic engagement can engender positive
4 For recent theoretical work on the origins of trust, see Carlin, Dorobantu, and Viswanathan
(2009).
5 The concept of trust is also related to the concept of integrity put forward in recent work by
Erhard, Jensen, and Zaffron (2009) and Erhard and Jensen (2015), who argue that trust follows
from a proactive stance to establish integrity—the process of honoring one’s word on commitments
made to a variety of constituents consistently.
6 The first two interpretations of social capital are often used in sociology and present social
capital as a resource for individuals built through networks (e.g., Coleman (1988), Lin (2001)); the
last two interpretations are often used in politics and economics and emphasize social capital as a
resource for facilitating cooperation at the group, community, or societal level.
outcomes by, for example, fostering trust and norms of cooperation, such as
reciprocity.7
Trust and cooperative norms comprise factors (social norms, including reci-
procity, and shared values) that shape the way that agents behave towards
each other and as members of society. Under this definition, social capital is
viewed as an enabler of collective action and cooperation, and thereby leading
to positive outcomes (e.g., economic growth, government performance, and en-
vironmental stewardship). The channels through which positive outcomes are
derived include: (i) reductions in transaction costs (by reducing the need for
formal contracts in the presence of information asymmetry (Knack and Keefer
(1997)) and (ii) potentially more efficient allocation of resources.
All of the above concepts are, of course, interconnected. For example, civic en-
gagement can generate trust and cooperation, which in turn can foster further
civic engagement; likewise, cooperation can build trust and vice versa. Further-
more, social capital can accrue at different levels, such as societal, institutional,
and individual levels. Hence, some individuals or institutions, including firms,
can invest more in social capital than others (see Coleman (1990), Leana and
Van Buren (1999), and Glaeser, Laibson, and Sacerdote (2002)).
B. Social Capital and Corporate Social Responsibility

To measure social capital at the firm level, we focus on a firm’s CSR activi-
ties. We motivate this metric by noting that definitions of CSR, which generally
involve aspects of civic engagement, shared beliefs, and disposition towards co-
operation between the firm and its stakeholders, tend to map directly into the
theoretical foundations of social capital. For example, one definition commonly
used by academics and practitioners, proposed by the World Business Coun-
cil for Sustainable Development (2000), is that “CSR is the commitment of a
business to contribute to sustainable economic development, working with em-
ployees, their families, the local community and society at large to improve the
quality of life.”
The belief that CSR activities can help build social capital and trust is
widespread among corporate managers. For example, in two recent CEO sur-
veys conducted by PricewaterhouseCoopers (2013, 2014), CEOs indicate having
plans to increase their firms’ engagement in CSR activities to restore stake-
holder trust after the crisis. In contrast, academic work linking social capital,
trust, and CSR is scarce but a recent book edited by Sacconi and Degli Antoni
(2011) presents a series of analytical studies showing that firms can build social
capital and trust through CSR investments.
Other recent work also supports the claim that CSR builds social capital and
enhances stakeholder trust in and cooperation with high-CSR firms. Eccles,
7 Putnam (1993, p. 172) defines the concept of generalized reciprocity as “a continuing rela-
tionship of exchange that is at a given time unrequited or imbalanced, but that involves mutual
expectations that a benefit granted out should be repaid in the future.” For earlier references to
reciprocity, see Gouldner (1960).
Ioannou, and Serafeim (2014) show that high-CSR firms implement processes
that consistently engage with stakeholders over the long term.8 Bénabou and
Tirole (2010) argue that stronger stakeholder engagement via CSR can lessen
the likelihood of short-term opportunistic behavior by managers, a view sup-
ported by empirical evidence in Gao, Lisic, and Zhang (2014) that executives of
high-CSR firms are less likely to engage in insider trading than executives of
low-CSR firms. In a similar vein, Kim, Park, and Wier (2012) find that socially
responsible firms are less likely to manage earnings.
While we acknowledge the limitations of CSR as an all-encompassing mea-
sure of firm-level social capital (see Scrivens and Smith (2013) and Sapienza,
Toldra-Simats, and Zingales (2013) for a discussion of social capital metrics), we
note that (i) CSR is measureable, albeit inexactly, (ii) CSR can have a nonnega-
tive payoff (see, e.g., Edmans (2011), Servaes and Tamayo (2013), and Flammer
(2015)), and (iii) firm-level CSR can change through investment or deprecia-
tion. Taken together, these three features alleviate Solow’s (1995) reservations
about social capital.9
C. Social Capital and Firm Valuation

In this paper, we argue that if a firm’s social capital helps build stakeholder
trust and cooperation, it should pay off more when being trustworthy is more
valuable, such as during an unexpectedly low-trust period.
From a shareholder perspective, Guiso, Sapienza, and Zingales (2008, p.
2557) posit that “the decision to invest in stocks requires not only an assessment
of the risk-return trade-off given the existing data, but also an act of faith (trust)
that the data in our possession are reliable and that the overall system is fair.”
During an unexpected decline in the general level of trust, outside shareholders
are likely to be more concerned that the financial information they previously
relied upon to guide investment decisions may not be credible. As such, they
will seek metrics such as social capital ratings that speak to a firm’s values and
integrity, placing a valuation premium on firms that are deemed to be more
trustworthy.
From the perspective of other stakeholders (e.g., employees, customers, sup-
pliers, and the community at large), much of their interaction with the firm
occurs through implicit or incomplete contracts, which may not be honored by
either party during a crisis. Social capital could facilitate these interactions
by fostering trust and cooperation (Putnam (1993)) and by reducing the need
for formal contracts (Knack and Keefer (1997)). For example, stakeholders
may perceive that the probability of breaching (implicit) contracts is lower for
high-social-capital firms due to shared values and cooperative norms. Likewise,
8 The idea of CSR as a competitive advantage is proposed and discussed in detail in Porter and
Kramer (2006, 2011).
9 Solow (1995, p. 38) argues that “if ‘social capital’ is to be more than a buzzword . . . There
needs to be an identifiable process of ‘investment’ that adds to the stock, and possibly a process
of ‘depreciation’ that subtracts from it. The stock of social capital should somehow be measurable,
even inexactly. Observable changes in it should correspond to investment and depreciation.”
stakeholders are more likely to “do whatever it takes” to help high-social-capital

firms weather a crisis, given that such firms displayed greater attention to, and
cooperation with stakeholders in the past.10 This observation is consistent with
the notion of reciprocity often discussed in studies of social capital and with
prior work showing that stakeholders tend to cooperate more when they per-
ceive firms to be trustworthy (e.g., for employees, see Guiso, Sapienza, and
Zingales (2015); for customers, see Servaes and Tamayo (2013)). The benefits
of social capital derived from stakeholder cooperation may be present during
any crisis, but as Sapienza and Zingales (2012) emphasize, cooperation breaks
down without trust. As such, firm-level social capital becomes even more rele-
vant when the level of trust in corporations, institutions, and capital markets
plummets, as occurred during the 2008–2009 financial crisis.
II. Sample and Summary Statistics

A. Sample Construction
To construct our sample, we gather information on firms’ CSR ratings from
the MSCI ESG Stats Database, which contains environmental, social, and
governance ratings of large publicly traded companies.11 This database con-
tains yearly ratings on roughly the 3,000 largest U.S. companies and has been
used in numerous studies examining the effect of CSR on firm performance
(e.g., Hong and Kostovetsky (2012), Deng, Kang, and Low (2013), Servaes and
Tamayo (2013), Krüger (2015), and Borisov, Goldman, and Gupta (2016)).12
ESG Stats classifies environmental, social, and governance performance into
13 different categories: community, diversity, employee relations, environment,
human rights, product, alcohol, gambling, firearms, military, nuclear, tobacco,
and corporate governance. As in Servaes and Tamayo (2013), we focus on the
first five of these categories. We do not include the product category in our
main analyses because it contains a number of elements that we consider to be
outside the scope of CSR, such as product quality and innovation; our findings
are unchanged, however, if we include the product category in our measure of
CSR. Similarly, we do not consider in our tests the ESG Stats categories that
10 Examples include business contacts continuing solid buying or selling relationships, em-
ployees working harder (or more cheaply) and more creatively to ensure success, or outside reg-
ulators/agencies being more sympathetic to these firms’ needs for direct relief or for flexibility
regarding regulations.
11 The MSCI ESG Stats database was previously known as the KLD Stats database, constructed
by Boston-based KLD Research and Analytics, Inc. (KLD).

12 The database is constructed as follows. MSCI first defines a number of ESG categories and
within each category, it specifies a number of criteria that capture good/poor ESG performance.
Once these criteria are set, MSCI scans public databases covering environmental issues, labor
issues, and the like to determine the ESG performance of the firm (e.g., has the firm committed
Environmental Protection Agency violations or had an industrial dispute?). Throughout this pro-
cess, MSCI assigns analysts to each firm to study the different elements of CSR, and relies on
sources beyond a firm’s reports or publicity regarding its green activities. See Krüger (2015) for a
more detailed description of the process MSCI follows to construct its CSR ratings.
penalize participation in the six industries that are considered controversial,

as there is nothing incremental that firms operating in these industries can do
to change their score except exit those industries (in addition, we control for
industry in all of our tests). Finally, we do not include the ESG Stats corporate
governance category in our main tests because governance is generally not part
of a firm’s CSR remit. However, as the governance category in aggregate, or
some of the individual governance category components, may be correlated with
the trustworthiness of a firm, we examine this category in robustness tests.
For each of the five categories we consider, ESG Stats compiles data on
both strengths and concerns. We are interested in capturing both elements; ac-
cordingly, we construct a net CSR measure that adds strengths and subtracts
concerns. As the maximum number of strengths and concerns for any given cat-
egory varies over time (e.g., the maximum number of strengths for community
is seven in 2005, but only four in 2010), we scale the strengths (concerns) for
each category by dividing the number of strengths (concerns) for each firm-year
by the maximum number of strengths (concerns) possible for that category in
that year. This procedure yields strength and concern indices that range from
zero to one for each category-year. Our measure of net CSR involvement in
each category-year is then obtained by subtracting the concerns index from
the strengths index. The net CSR index per category therefore ranges from –1
to +1. Finally, to obtain our primary explanatory variable, a firm’s total net
CSR index (CSR hereafter), we combine the net CSR indices for the categories
of community, diversity, employee relations, environment, and human rights.
This is a net measure across our set of stakeholder-oriented categories, and
ranges from –5 to +5. There is substantial variation in this measure across
firms and industries. For example, in 2006, in the Apparel Retail Industry,
GAP has a score of 0.40 while Limited Brands scores –0.53; in Chemicals, the
score for Air Products and Chemicals is 0.16, while that for Celanese is –1.36.13
We obtain stock return data from CRSP and accounting data from Compu-
stat. We remove financial firms from our sample due to the extensive amount
of government support given to such firms during the crisis. We also remove
micro-cap stocks (those with a market capitalization below $250 million as of
year-end 2007) because these stocks tend to have low liquidity and high bid-ask
spreads, and are subject to more price pressure effects of trading, all of which
would likely be more pronounced during the financial crisis.
As in Lins, Volpin, and Wagner (2013), we define the financial crisis as the
period from August 2008 to March 2009. August of 2008 preceded the Septem-
ber 2008 Lehman Brothers bankruptcy, while March of 2009 is when the S&P
500 hit its lowest point of the crisis. This period also corresponds to the time of
a severe decline in trust as suggested by Sapienza and Zingales (2012) (see also
13 Limited Brands has employee, human rights, and diversity concerns that are only partially
outweighed by diversity strengths; GAP, in contrast, has strengths in diversity, human rights, and
community, with some employee concerns. Celanese’s score is due to environmental, employee, and
diversity concerns; Air Products & Chemicals also has environmental and employee concerns, but
its strengths in these areas outweigh the concerns.
Tonkiss (2009)). The decline in trust later in 2008 is also corroborated by the
Trust Barometer developed by Edelman, the world’s largest independent public
relations firm, which conducts global surveys of trust in business, government,
NGOs, and the media—they report that trust in business in the United States
declined from 58% in early 2008 to 38% in early 2009.14
The main stock return measures for each firm are Raw Crisis-Period Return,
which is the firm’s raw buy-and-hold return from August 2008 through March
2009, and Abnormal Crisis-Period Return, which is the raw return minus the
expected return, based on the market model estimated over the 60-month pe-
riod ending in July 2008. To avoid problems with outliers, we winsorize these
returns at the 1st and 99th percentiles. We relate these return measures to our
CSR measure for the year 2006 to guard against the possibility that by year-
end 2007 firms may have already changed their CSR policies in anticipation of
the crisis ahead.15
After combining nonfinancial firms with sufficient data coverage on the CRSP
and Compustat databases and firms on the ESG Stats database, we obtain a
sample of 1,673 nonfinancial firms for which all explanatory variables are
available for the crisis period.
B. Descriptive Statistics
Table I provides descriptive statistics for our main variables. The first row of
Panel A shows that our primary variable of interest, CSR, is slightly negative
with a mean value of –0.165 and a median value of –0.200. Thus, the average
and median firm has more CSR concerns than strengths, consistent with Deng,
Kang, and Low (2013), Servaes and Tamayo (2013), and Borisov, Goldman, and
Gupta (2016). The next row shows that Raw Crisis-Period Return is strongly
negative, with a mean of –39.1%, a median of –40.3%, and a 25th percentile
value of –59.5%, indicating that investors and other stakeholders were likely
quite concerned about the survival prospects of many of the firms they held
in their portfolios, worked for, or interacted with in business transactions. The
median abnormal return is close to zero at 1.3%, while the mean is 11.6%. Panel
A also provides definitions and descriptive statistics for firm characteristics
that we use as control variables in our models; Panel B presents a correlation
matrix of all the variables employed in our main analyses.
14 The Global Competitiveness Index developed by the World Economic Forum, and based in
part on survey evidence, also contains a component measuring trust in financial markets. Released
in September of each year, this trust measure shows a decline from 5.65 in September 2008 to 5.06
in September 2009. The Financial Trust Index (financialtrustindex.org) developed by Sapienza
and Zingales is another measure of public trust, but because this index started after the onset of
the crisis we cannot employ it to corroborate the extent to which trust changed as a result of the
crisis.
15 We repeated all of our analyses using CSR measured at year-end 2005 in case 2006 CSR data
partially reflect anticipation of a future crisis. All of our findings continue to hold. These results
are reported in the Internet Appendix, available in the online version of the article on The Journal
of Finance website.
Table I
Descriptive Statistics
The sample consists of 1,673 firms with CSR data available from the MSCI ESG STATS database as of year-end 2006 and returns available during the
period August 2008 to March 2009. CSR is the total net (strengths minus concerns) CSR score computed using five stakeholder-oriented categories
(environment, employee relations, human rights, community, and diversity). To compute the total net CSR measure, we first compute the net CSR
index within each of the categories. The net CSR index for each category is computed by taking the number of strengths identified for a given firm
and dividing this by the maximum possible strengths in that category, and then subtracting the number of concerns identified for the firm divided by
the maximum possible concerns; the net CSR score for each category thus ranges from –1 to +1. The total net CSR measure, CSR, is computed as the
sum of the net CSR indices for the five categories and ranges from –5 to +5. CSR ratings are measured at the end of 2006. Crisis-Period Raw Return is
the raw return computed over the period August 2008 to March 2009. Crisis-Period Abn. Return is the market model-adjusted return over the period
August 2008 to March 2009, with market model parameters computed over the five-year period ending in July 2008 using the CRSP value-weighted
index as the market proxy. Accounting data are based on the last quarter ending at or before the end of 2007. Market Capitalization is in millions
of dollars. Long-Term Debt is computed as long-term debt divided by assets. Short-Term Debt is computed as debt in current liabilities divided by
assets. Cash Holdings is computed as cash and marketable securities divided by assets. Profitability is computed as operating income divided by
assets. Book-to-Market is computed as book value of equity divided by market value of equity. Negative B/M is a dummy variable set to one when
the book-to-market ratio is negative and zero otherwise. Momentum is the raw return over the period August 2007 to July 2008. Idiosyncratic Risk
is computed as the residual variance from the market model estimated over the five-year period ending in July 2008, using monthly data. Financial
firms and micro-cap firms, which we define as firms with a market capitalization below $250 million as of year-end 2007, are removed from the
sample. Control variables and returns are winsorized at the 1st and 99th percentiles.
Panel A: Summary Statistics
Mean SD (Std Dev) 25th perc. Median 75th perc.
CSR −0.165 0.381 −0.343 −0.200 0.006

Crisis-Period Raw Return −0.391 0.284 −0.595 −0.403 −0.211
Crisis-Period Abn. Return 0.116 0.592 −0.275 0.013 0.383
Market Capitalization 6922 23941 598 1327 4010
Long-Term Debt 0.198 0.193 0.011 0.170 0.307
Social Capital, Trust, and Firm Performance
Short-Term Debt 0.029 0.055 0 0.0055 0.031

Cash Holdings 0.172 0.199 0.026 0.088 0.247
Profitability 0.033 0.034 0.021 0.034 0.049
Book-to-Market 0.430 0.295 0.231 0.377 0.576
Negative B/M 0 0.155 0 0 0
Momentum −0.082 0.370 −0.322 −0.110 0.116
Idiosyncratic Risk 0.011 0.010 0.005 0.009 0.015
1795
(Continued)
1796
Table I—Continued
Panel B: Correlation Matrix
Crisis Crisis
Raw Abn. Ln (Mkt Cash
CSR Return Return Cap) L/T Debt S/T Debt Hold. Profit. B/M Neg. B/M Mom.
Crisis Raw Return 0.11

Crisis Abn. Return 0.08 0.72
Ln(Mkt Cap) 0.20 0.09 −0.09
Long-Term Debt −0.07 −0.10 −0.10 0.05
Short-Term Debt 0.06 −0.00 −0.04 0.11 0.01
Cash Holdings 0.06 0.10 0.24 −0.20 −0.33 −0.11
Profitability 0.05 0.06 −0.06 0.24 −0.05 −0.03 −0.30
Book-to-Market 0.01
The Journal of FinanceR
−0.09 −0.10 −0.02 −0.25 −0.11 −0.21 −0.19

Negative B/M −0.02 −0.01 0.02 −0.06 0.38 0.06 0.04 −0.01 −0.30
Momentum −0.08 −0.03 −0.35 0.14 −0.09 −0.01 −0.03 0.13 −0.22 −0.04
Idiosyncratic Risk −0.12 −0.13 0.11 −0.39 −0.03 −0.07 0.41 −0.31 −0.11 0.09 −0.08
III. Crisis-Period Returns

A. Baseline Results
We estimate various regression models of stock returns during the crisis
period as a function of firms’ pre-crisis CSR ratings and a number of control
variables. Panel A of Table II contains our baseline regression models. The
dependent variable in columns (1) and (3) is Raw Crisis-Period Return, while
in columns (2) and (4), it is Abnormal Crisis-Period Return. Our variable of
interest is the firm’s CSR measured at year-end 2006. In all models, we include
industry dummies (defined at the two-digit SIC level) because some industries
may be more likely to invest in CSR than others and may have been differ-
entially affected by the financial crisis. We also control for the firm’s factor
loadings based on the Fama-French three-factor model plus the momentum
factor.16
Columns (1) and (2) show that firms with higher CSR ratings performed
significantly better during the crisis. The effect of CSR on returns is economi-
cally large: a one-standard-deviation increase in CSR (0.381) is associated with
a 2.25 percentage point increase in raw returns and a 4.15 percentage point
increase in abnormal returns during the crisis.
One concern with the specifications reported in columns (1) and (2) is that the
strong performance of high-CSR firms during the crisis may be due to omitted
variables that happen to be correlated with CSR, rather than due to CSR itself.
To address this possibility, in columns (3) and (4), we control for a firm’s financial
health in the year before the crisis and for other firm characteristics that have
been found to affect stock returns. We employ several proxies to measure a
firm’s financial health and, thus, its ability to withstand a severe downturn in
the economy: Cash Holdings (cash and marketable securities divided by assets),
Short-Term Debt (debt in current liabilities divided by assets), Long-Term Debt
(long-term debt divided by assets), and Profitability (operating income divided
by assets). During a crisis, profitable, cash-rich firms with low debt can continue
investing, while other firms may be forced to cut investment, especially if
they have short-term debt maturing during the crisis (see Duchin, Ozbas, and
Sensoy (2010), Almeida et al. (2012), and Harford, Klasa, and Maxwell (2014)
for empirical evidence consistent with these expectations).
Because additional firm characteristics may also be important for stock re-
turns (see, e.g., Daniel and Titman (1997)), we also control for Size (the log
of a firm’s equity market capitalization), Book-to-Market (book value of equity
divided by market value of equity), and Momentum (the firm’s raw return over
the period August 2007 to July 2008). We also add a dummy for firms with a
Negative Book-to-Market ratio, as such firms are likely distressed and hence
their returns may behave more like those of high book-to-market firms than
low book-to-market firms (see Fama and French (1992)). Finally, we control
16 We estimate the factor loadings over the 60 months prior to the onset of the crisis, using factor
returns obtained from Kenneth French’s website. Firms are excluded from the analysis if fewer
than 12 months of data are available to estimate factor loadings.
Table II
Crisis-Period Returns and CSR
This table presents regression estimates of crisis-period returns on CSR and control variables.
Crisis-period returns are measured as both raw buy and hold returns and abnormal returns over
the period August 2008 to March 2009. In Panel A, we use a linear measure of CSR, CSR, which is
the net (strengths minus concerns) CSR score computed using five stakeholder-oriented categories,
measured at the end of 2006. In Panel B, we use dummy variables for CSR quartiles such that
CSR2 takes the value of one if the firm is in the second CSR quartile and zero otherwise, CSR3
takes the value of one if the firm is in the third CSR quartile and zero otherwise, and CSR4 takes
the value of one if the firm is in the fourth CSR quartile and zero otherwise. In Panel C, we employ
the linear measure of CSR and add measures of corporate governance. ESG Stats Governance
Index is the net CSR index for the governance category and is computed by taking the number
of strengths, divided by the possible maximum, and subtracting the number of concerns, divided
by the possible maximum; this measure thus ranges from –1 to +1. The E-Index is the sum of
six dummies reflecting the following antitakeover provisions: (i) a staggered board, (ii) limits to
amend the charter, (iii) limits to amend bylaws, (iv) supermajority voting requirements, (v) golden
parachutes for executives, and (vi) the ability to adopt a poison pill (see Bebchuk, Cohen, and
Ferrell (2009)), obtained from MSCI Governance Metrics. Board Independence (fraction of board
consisting of outside directors), Board Size, a dummy if the CEO Is Not the Chairman, and Board
Ownership (fraction of outstanding shares owned by board members) are obtained from the MSCI
Directors database. When the governance metrics are not available on the MSCI databases, we
set them to zero and code a missing variable dummy that we set to one if that governance item
is missing. These dummies are included in all models, but their coefficients are not reported. The
control variables are as defined in Table I. Industry dummies are defined at the two-digit SIC code
level. Financial firms and micro-cap firms with a market capitalization below $250 million are
removed from the sample. The control variables and returns are winsorized at the 1st and 99th
percentiles. Heteroskedasticity-consistent standard errors are presented in parentheses. ***, **,
and * indicate that the parameter estimate is significantly different from zero at the 1%, 5%, and
10% level, respectively.
Panel A: Net CSR Score: Raw and Abnormal Returns
Raw return Abnormal return Raw return Abnormal return

(1) (2) (3) (4)
CSR 0.059*** 0.109*** 0.048*** 0.087***

(0.018) (0.027) (0.017) (0.032)
Ln(Market Cap) 0.001 −0.015
(0.005) (0.011)
Long-Term Debt −0.112*** −0.102
(0.046) (0.086)
Short-Term Debt −0.323*** −0.384*
(0.115) (0.219)
Cash Holdings 0.175*** 0.380***
(0.047) (0.091)
Profitability 0.528** 0.732
(0.261) (0.509)
Book-to-Market −0.116*** −0.045
(0.030) (0.058)
Negative B/M −0.015 0.049
(0.061) (0.127)
Momentum −0.030 −0.285
(0.024) (0.044)
(Continued)
Table II—Continued
Panel A: Net CSR Score: Raw and Abnormal Returns

(1) (2) (3) (4)
Idiosyncratic Risk −3.155*** −8.870***

(0.876) (1.719)
Constant −0.588*** −0.781*** −0.528*** 0.0867***
(0.260) (0.035) (0.059) (0.032)
Four-factor loadings Yes Yes Yes Yes
Industry dummies Yes Yes Yes Yes
N 1,673 1,673 1,673 1,673
Adj. R2 0.17 0.33 0.20 0.37
Panel B: Dummies for Quartiles of Net CSR Score: Raw and Abnormal Returns

(1) (2) (3) (4)
CSR2 0.0303 0.0560 0.0296 0.0480

(0.0192) (0.0347) (0.0188) (0.0341)
CSR3 0.0365* 0.0649* 0.0405** 0.0562
(0.0197) (0.0373) (0.0192) (0.0361)
CSR4 0.0552*** 0.0985*** 0.0453** 0.0727**
(0.0126) (0.0356) (0.0194) (0.0362)
Ln(Market Cap) 0.002 −0.012
(0.006) (0.011)
Long-Term Debt −0.113*** −0.103
(0.046) (0.087)
Short-Term Debt −0.328*** −0.386*
(0.115) (0.219)
Cash Holdings 0.175*** 0.381***
(0.047) (0.091)
Profitability 0.530** 0.734
(0.260) (0.508)
Book-to-Market −0.117*** −0.046
(0.030) (0.057)
Negative B/M −0.017 0.046
(0.061) (0.058)
Momentum −0.031 −0.288
(0.024) (0.044)
Idiosyncratic Risk −3.184*** −8.934***
(0.874) (1.729)
N 1,673 1,673 1,673 1,673
Adj. R2 0.16 0.32 0.20 0.36
(Continued)
Table II—Continued
Panel C: Controlling for Corporate Governance

(1) (2) (3) (4)
CSR 0.048*** 0.090*** 0.049*** 0.088***

(0.017) (0.032) (0.017) (0.032)
ESG Stats Governance Index 0.008 −0.151
(0.050) (0.092)
E-Index −0.008* −0.018**
(0.005) (0.009)
Board Independence −0.080 −0.085
(0.072) (0.140)
Board Size 0.002 −0.000
(0.004) (0.007)
CEO Is Not Chair −0.010 −0.015
(0.015) (0.026)
Board Ownership −0.039 −0.034
(0.074) (0.147)
Firm characteristics Yes Yes Yes Yes
N 1,673 1,673 1,673 1,673
Adj. R2 0.20 0.37 0.20 0.37
for a firm’s Idiosyncratic Risk (the residual variance from the market model
estimated over the five-year period ending in July 2008, using monthly data)
under the premise that stock price volatility may also affect returns (Goyal
and Santa-Clara (2003)). We measure financial health and firm characteristics
at the end of December of 2007, or as close as possible to it for firms that do
not have a December fiscal year-end, except for momentum and idiosyncratic
volatility, which are computed over one and five years, respectively, before the
start of the crisis period.
The results presented in columns (3) and (4) of Panel A of Table II confirm
that high-CSR firms had higher stock returns during the crisis. The magni-
tude of the high-CSR outperformance is somewhat attenuated after we include
additional control variables, but the effect is still economically important. For
example, in the model in column (3), a one-standard-deviation increase in CSR
(0.381) is associated with a 1.83 percentage point increase in raw crisis-period
returns.
Turning to the control variables, as expected, firms that entered the crisis in
better financial health (higher cash holdings and profitability and lower debt)
have higher crisis-period stock returns, while firms with higher idiosyncratic
risk had lower returns. In terms of economic significance, the effects of leverage,
cash holdings, and idiosyncratic risk are the largest. Based on the model in
column (3), a one-standard-deviation increase in long-term debt (0.19), cash
holdings (0.20), and idiosyncratic risk (0.01) is associated with a change in raw
crisis-period returns of –2.16, 3.48, and –3.16 percentage points, respectively.
Thus, the economic impact of CSR ratings on returns during the crisis is more
than four-fifths of the impact of leverage and more than half of the impact of
cash holdings and volatility, indicating that social capital is indeed important
in explaining crisis-period returns.
In Panel B of Table II, we re-estimate our previous models, but instead of
including our linear measure of CSR as an explanatory variable, we divide
firms into CSR quartiles and include dummies for quartiles 2–4 (the intercept
captures the effect of quartile 1). This approach allows us to assess whether the
effect of a firm’s social capital on returns is more pronounced at very high or
very low levels of social capital. The results again show that firms with better
CSR ratings had the largest crisis-period returns. The difference in raw returns
between firms in the best and worst CSR quartiles, as captured by the coef-
ficient on CSR4, is 5.52 percentage points when we omit firm characteristics
and 4.53 percentage points when we include them. For abnormal returns, the
difference is even greater at 9.85 percentage points and 7.27 percentage points,
respectively. The impact of CSR on returns is monotonic, but not entirely lin-
ear. Based on the model in column (4), which features the full set of control
variables, abnormal returns increase about 4.8 percentage points when moving
from the lowest to the 2nd lowest quartile of CSR. Only modest improvements
in returns accrue when moving to the 3rd quartile, while a move from the 3rd
to the 4th quartile yields a more significant improvement in returns of 1.65
percentage points. These results indicate that investors were most concerned
when a firm had a low level of social capital and most reassured when firm
social capital was high.
We also ensure that our findings persist after we control for measures of
corporate governance. Recent evidence shows that better governed firms per-
formed relatively well during the financial crisis (Lins, Volpin, and Wagner
(2013) and Nguyen, Nguyen, and Yin (2015)). If governance is correlated with
our CSR measure, then it is possible that CSR is simply proxying for gover-
nance, resulting in an omitted variable bias. To address this concern, we gather
data on a variety of governance measures as of year-end 2006. We first use our
prior method to construct a governance measure from the ESG Stats database:
for each firm, the number of governance concerns is divided by its possible
maximum and subtracted from the number of strengths divided by its possible
maximum, yielding a governance index that ranges from –1 to +1. We also
measure governance using the firm’s E-Index (the entrenchment index featur-
ing the six governance provisions17 identified in Bebchuk, Cohen, and Ferrell
(2009)), Board Independence (the fraction of the board consisting of outside
directors), Board Size, a dummy if the CEO Is Not the Chairman, and Board
Ownership (the fraction of outstanding shares owned by the board members),
obtained from the MSCI Governance Metrics and Directors databases.
17 The E-index consists of the following six governance provisions that indicate entrenchment:
a staggered board, limits to amend the charter, limits to amend bylaws, supermajority voting
requirements, golden parachutes for executives, and the ability to adopt a poison pill (see Bebchuk,
Cohen, and Ferrell (2009)).
In Panel C of Table II, we repeat the analyses from Panel A, but we now
add the governance controls. All models include the full set of other control
variables employed in Panels A and B. Columns (1) and (2) show that the
ESG Stats Governance Index is not significantly related to raw or abnormal
crisis-period returns and that the impact of CSR on crisis-period returns is
virtually identical to that reported previously. This evidence suggests that the
CSR effect is not picking up a governance component. The models in columns
(3) and (4) include all other governance measures. We again find that the effect
of CSR on crisis-period returns persists. The E-Index is significant for both raw
and abnormal return models, which indicates that firms with more entrenched
managers performed worse during the crisis. The other governance provisions
are insignificant.
To get a sense of the costs associated with firms’ CSR activities, we follow
Di Giuli and Kostovetsky (2014) and estimate a regression model (reported
in the Internet Appendix) of the log of Selling, General, and Administrative
(SG&A) expenses measured in 2006 as a function of CSR and a number of
control variables (log assets, equity book-to-market, cash holdings to assets,
total interest bearing debt to assets, dividend payments to assets, and income
before extraordinary items to assets). Increasing CSR from its 1st to its 4th
quartile is associated with SG&A expenses that are $44.9 million higher for
the median firm in our sample and $203.5 million for the mean firm. These
cost estimates are substantial and may help explain why not all firms choose
to engage in CSR activities.
Overall, the findings reported in Table II show that more socially responsible
firms suffered less during the crisis, and that this effect is not due to differences
in financial strength or corporate governance.18 These results are consistent
with the view that firm investments in social capital provided investors a
greater sense of trust in the firm as the crisis unfolded, leading to relative
stock price outperformance.
B. Excess Returns and CSR during the Enron/Worldcom Fraud Scandals

The above findings provide evidence of a positive relation between CSR and
excess returns during the 2008–2009 crisis period, when the overall level of
trust in corporations suffered a severe shock. As this crisis was arguably
the most severe crisis of confidence in generations, few other economy-wide
shocks to trust could have the same effect. Perhaps one “shock” that comes
18 We also examine whether CSR is proxying for reporting transparency or (the lack of) ac-
counting concerns, using the following ESG Stats measures: CSR transparency strength, CSR
transparency concern, and accounting concern. We find no evidence that this is the case. We also
include an indicator variable set to one if the firm is included in the 100 Best Companies to Work For
list as published by Fortune Magazine in February 2008. Edmans (2011) finds that firms included
in this list earn excess returns over the subsequent five-year period. This dummy is not significant
in explaining crisis-period returns and its inclusion does not have any impact on the coefficient on
our measure of CSR.
close to a general crisis of trust is the one caused by the ripple effects as-
sociated with the frauds and subsequent bankruptcies of Enron, Worldcom,
and several other large firms. Enron filed for bankruptcy in December of
2001, after admitting to accounting violations in October 2001. Of course,
fraud committed by one company alone does not necessarily dent trust in
all firms, but soon after Enron’s accounting violations were revealed other
cases came into the spotlight. In the last quarter of 2001 and the first half of
2002 alone, Adelphia, Bristol-Myers Squibb, Global Crossing, Homestore.com,
ImClone Systems, Kmart, Qwest, Tyco, and Worldcom were all in the news be-
cause of accounting irregularities and/or outright fraud. Global Crossing, Adel-
phia, and Worldcom filed for bankruptcy in January, June, and July of 2002,
respectively.
These bankruptcies and scandals are likely to have caused a general decline
in trust in corporations. Several newspaper articles published at the time sup-
port this belief. For example, on July 14, 2002, Associated Press Newswires
published an article entitled: “How much will the loss of trust in U.S. busi-
nesses hurt the economy?” and on July 17, The Wall Street Journal reported on
a warning from Alan Greenspan, the Federal Reserve Chairman at the time,
that breakdowns in corporate governance could undermine the trust necessary
for efficient markets. On December 31, 2002, the Financial Times, in discussing
the accuracy of forecasts made for 2002, stated that: “ . . . even fewer divined
that a loss of trust in company statements would be the trigger for another
growl of the bear market.”
To determine whether our findings also hold around the Enron crisis, we
follow the same procedure as the one employed for the 2008–2009 finan-
cial crisis. Specifically, we cumulate returns over the period October 2001,
when the Enron accounting violations were first revealed, to March 2003, the
month prior to the beginning of the stock market rally that persisted until
the start of the 2008–2009 crisis. We relate these returns to CSR computed
as of year-end 2000 and the same control variables as employed previously.
Small firms (market value below $250 million in 2007 dollars) are again ex-
cluded, but this time, we include financial firms as they were not uniquely
supported by the government during this period. Factor loadings are com-
puted over the five-year period ending in September 2001. Unfortunately, ESG
Stats coverage for this period is much smaller, yielding a sample of only 412
companies.
Table III contains the results. We report two specifications each for raw and
abnormal returns. The models in columns (1) and (2) use our primary measure
of CSR (the difference between scaled CSR strengths and weaknesses) as the
key explanatory variable, while the models in columns (3) and (4) contain a
coarser CSR measure, namely, a dummy variable set to one if CSR is positive
and zero otherwise. The latter specification may be better suited in this case,
given the small sample size and the nonlinearities in the CSR-return relation
reported in Panel B of Table II.
As illustrated in both columns (1) and (2), the coefficient on our linear
measure of CSR, while positive, is not significantly different from zero. The
Table III
CSR and Returns during the Enron/Worldcom Fraud Scandals
This table presents regression estimates of raw and abnormal stock returns from October 2001
to March 2003 as a function of CSR. Abnormal returns are computed based on the market model
using the CRSP value-weighted index as the market proxy. Market model parameters are estimated
using monthly data over the five-year period ending in September 2001. CSR ratings are measured
at the end of 2000. The control variables are the same as those employed in Table II. All financial
controls are measured at year-end 2000 or as close to it as possible for firms without December fiscal
year-ends. Fama-French and momentum factor loadings are computed using monthly data over the
five-year period ending in September 2001. The firm characteristics employed as control variables
are the same as those in Table II, except that they are measured as of year-end 2000. Industry
dummies are defined at the two-digit SIC code level. Micro-cap firms with a market capitalization
below $250 million (in 2007 dollars) are removed from the sample. The control variables and returns
are winsorized at the 1st and 99th percentiles. Heteroskedasticity-consistent standard errors are
presented in parentheses. ***, **, and * indicate that the parameter estimate is significantly
different from zero at the 1%, 5%, and 10% level, respectively.
Raw Return Abnormal Return Raw Return Abnormal Return

(1) (2) (3) (4)
Constant 1.058*** 0.977*** 0.116 0.244

(0.176) (0.208) (0.219) (0.267)
CSR 0.026 0.060
(0.044) (0.053)
CSR > 0 indicator 0.073* 0.095**
(0.039) (0.044)
Firm characteristics Yes Yes Yes Yes
N 412 412 412 412
Adj. R2 0.16 0.20 0.17 0.21
indicator variable for high-CSR firms, which is employed as an explanatory

variable in the models in columns (3) and (4), is significant, and indicates that
firms with positive CSR scores had 7.3 percentage points higher raw returns
(9.5 percentage points higher abnormal returns) than firms with negative CSR
scores. This finding suggests that, during the crisis of confidence surround-
ing the accounting scandals revealed in 2001 and 2002, high-CSR firms again
earned excess returns relative to low-CSR companies.
We confirm that our findings are robust to different starting and ending
points of this crisis. Abnormal returns are positively related to the high-CSR
dummy (at the 10% significance level or better) for any starting month between
October 2001 (the month of the Enron bankruptcy) and May 2002 (after many
other firms had revealed accounting irregularities), and for any ending month
between December 2002 and March 2003 (when the stock market recovery
started).
These results support our prior findings that social capital created through
CSR activities matters more when investor confidence in corporations has been
damaged.
C. Comparing Returns Inside and Outside of the Crisis Period

Our evidence so far indicates that CSR positively affected stock returns dur-
ing two periods when overall trust in corporations, institutions, and financial
markets declined. In this section, we investigate whether this positive relation
is unique to periods of low trust or is common to most periods, perhaps due to
some unobservable (omitted) risk factor that is correlated with CSR.
To address this question, we estimate a difference-in-differences model with
continuous treatment and include firm and time fixed effects. Specifically, we
construct a panel of monthly returns for all the firms in our sample starting in
2007, prior to the onset of the crisis, and ending in 2013, several years into the
economic recovery. For this panel, we estimate the following model:19
Returni,t = bo + b1 C SRi,2006 × Crisist + b2 C SRi,2006 × Post-Crisist + b3 Xi,t−1
+ T ime Dummies + Firm Fixed Effects + ei,t , (1)
where Returni,t is the monthly raw or market-model adjusted return, CSRi,2006
is our proxy for CSR, measured at year-end 2006, Crisist is a dummy variable
set to one in the period August 2008 to March 2009, Post-Crisist is a dummy
variable set to one in the period April 2009 to December 2013, and Xi,t–1 is a vec-
tor of control variables. The control variables are the firm financial characteris-
tics and factor loadings employed in Table II, but updated annually (accounting
variables) or monthly (market-based variables). We measure CSR 20 months
before the onset of the crisis to eliminate any concern that firms adjusted their
CSR policies in anticipation of the crisis. To ensure that the accounting data
are publicly available, we leave a three-month gap after the fiscal year-end
to update the data. Factor loadings are re-estimated each month based on the
previous 60 months’ data. Time dummies are specified at the monthly level and
firm fixed effects control for time-invariant omitted risk factors. The firm’s CSR
itself is absorbed by the firm fixed effects. All standard errors are clustered at
the firm level.20 As in Table II, firms with market values below $250 million as
of year-end 2007 are excluded from the analysis. The coefficient on the inter-
action between 2006 CSR and the crisis (b1 ) captures the differential impact
of CSR on monthly stock returns during the eight-month period from August
2008 to March 2009, after controlling for the firm’s factor loadings and financial
characteristics and after removing both the firm’s average return (firm fixed
effects) over the entire estimation period and any time-series pattern in overall
returns (time fixed effects).
The results for both raw and market-model adjusted returns are presented
in Panel A of Table IV. Both specifications indicate that high-CSR firms exhibit
superior performance during the crisis period; after the crisis, the relation
between CSR and returns becomes insignificant. In terms of economic signifi-
cance, the coefficient of 0.0201 on the Crisis × CSR interaction indicates that
19 We are grateful to an anonymous referee for suggesting this test.

20 Significance levels are virtually identical if we double cluster standard errors by firm and
time period.
Table IV
Abnormal Returns Surrounding the Crisis and CSR
This table presents results of estimating the following panel regression model:
Returni,t = bo + b1 C SRi,2006 × Crisist + b2 C SRi,2006 × Post-Crisist + b3 Xi,t−1

+ T ime Dummies + Firm Fixed Effects + ei,t ,
where Returni,t is the monthly raw or market-model adjusted return, CSRi,2006 is our proxy for
CSR, measured at year-end 2006, Crisist is a dummy variable set to one in the period August 2008
to March 2009, Post-Crisist is a dummy variable set to one in the period April 2009 to December
2013, and Xi,t–1 is a vector of control variables. Panel A reports results using the overall measure
of CSR. In Panel B, we create an additional interaction between CSR and the period July 2007
to July 2008, when there was a shock to the supply of credit. In Panel C, we allow the effect of
CSR on returns to depend on whether the firm is headquartered in a low-trust or high-trust region
based on the 2006 General Social Survey. In Panel D, CSR is split into two components: Internal
Stakeholder CSR, which combines the measures for diversity and employee relations, and External
Stakeholder CSR, which combines the measures for community, environment, and human rights.
The control variables are the same as those employed in Table II and they include firm financial
characteristics as well as factor loadings. The financial characteristics based on accounting data
are updated three months after each fiscal year-end. The characteristics based on market data
(momentum, size, market-to-book, factor loadings) are updated monthly. Factor loadings are re-
estimated each month based on the previous 60 months’ data. The regression is estimated over the
period 2007–2013. Financial firms and micro-cap firms with a market capitalization below $250
million (in 2007 dollars) are removed from the sample. Except when otherwise indicated, numbers
in parentheses are heteroskedasticity-consistent standard errors, clustered at the firm level. ***,
**, and * indicate that the parameter estimate is significantly different from zero at the 1%, 5%,
and 10% level, respectively.
Panel A: Overall CSR
Variable Raw return Abnormal return
CSR × Crisis 0.0201*** 0.0153***

(0.0037) (0.0039)
CSR × Post-Crisis 0.0018 0.0020
(0.0022) (0.0024)
Firm characteristics Yes Yes
Four-factor loadings Yes Yes
Firm fixed effects Yes Yes
Time (monthly) fixed effects Yes Yes
Standard errors clustered by Firm Firm
CSR × (Crisis – Post-Crisis) 0.0183 0.0124
p-Value (0.00) (0.00)
N 121,247 121,247
Adj. R2 0.29 0.06
Panel B: CSR during the Shock to the Supply of Credit and the Crisis
CSR × Shock to Credit 0.0021 0.0001

(0.0030) (0.0030)
CSR × Crisis 0.0216*** 0.0154***
(0.0041) (0.0040)
(Continued)
Table IV—Continued
Panel B: CSR during the Shock to the Supply of Credit and the Crisis
CSR × Post-Crisis 0.0033 0.0021

(0.0027) (0.0028)
CSR × (Crisis – Shock to Credit) 0.0195 0.0153
p-Value (0.00) (0.00)
CSR × (Post-Crisis – Crisis) −0.0183 −0.0133
p-Value (0.00) (0.00)
N 121,247 121,247
Adj. R2 0.29 0.06
Panel C: Split by High- and Low-Trust Regions
CSR × Crisis × High Trust 0.0275*** 0.0171**

(0.0073) (0.0074)
CSR × Crisis × Low Trust 0.0154*** 0.0130***
(0.0047) (0.0047)
CSR × Post-Crisis × High Trust −0.0079* −0.0048
(0.0047) (0.0050)
CSR × Post-Crisis × Low Trust 0.0037 0.0025
(0.0024) (0.0027)
CSR × (Crisis – Post-Crisis) × High Trust 0.0354 0.0219
p-Value (0.00) (0.00)
CSR × (Crisis – Post-Crisis) × Low Trust 0.0117 0.0105
p-Value (0.01) (0.01)
CSR × Crisis × (High – Low Trust) 0.0121 0.0041
p-Value (0.16) (0.64)
N 115,453 115,453
Adj. R2 0.29 0.07
Panel D: Internal and External Stakeholder CSR
Int. Stakeholder CSR × Crisis 0.0225*** 0.0140***

(0.0048) (0.0049)
Int. Stakeholder CSR × Post-Crisis 0.0002 0.0049**
(0.0027) (0.0029)
Ext. Stakeholder CSR × Crisis 0.0144** 0.0185***
(0.0072) (0.0071)
(Continued)
Table IV—Continued
Panel D: Internal and External Stakeholder CSR
Ext. Stakeholder CSR × Post-Crisis 0.0057* −0.0048

(0.0040) (0.0044)
Int. Stakeholder CSR × (Crisis – Post-Crisis) 0.0223 0.0091
p-Value (0.00) (0.04)
Ext. Stakeholder CSR × (Crisis – Post-Crisis) 0.0087 0.0233
p-Value (0.20) (0.00)
N 121,247 121,247
Adj. R2 0.29 0.06
a one-standard-deviation increase in 2006 CSR (0.381) is associated with a

77-basis-point higher return during the crisis on a monthly basis.21,22 These
results indicate that the excess returns earned by high-CSR firms are limited
to the crisis period, consistent with our suggestion that social capital created
through CSR pays off when trust in the economy declines unexpectedly.
The lack of a reversal in abnormal returns for high-CSR firms after the
crisis may appear surprising. Such a reversal rests on the assumption that
overall trust in firms and markets has fully recovered. Trust has remained
relatively low since the crisis, however. For example, according to the Financial
Trust Index, 11% of respondents trusted the stock market and 12% trusted
large corporations in December 2009. These figures increased to 13% and 16%,
respectively, by December 2012, but they are still suggestive of a low level of
trust following the crisis.23 This would be consistent with the lack of return
reversals. We also note that high-trust firms should not earn further positive
21 We also estimated this model without the monthly time dummies, but with dummies for the
crisis period and the post-crisis period. These dummies capture the change in returns during and
after the crisis for firms with a CSR score of zero. The coefficient on the crisis dummy indicates
an average monthly decline during the crisis of 7.56 percentage points relative to the pre-crisis
period. In the post-crisis period, the raw returns are 1.60 percentage points higher per month than
in the pre-crisis period (see the Internet Appendix).
22 Gormley and Matsa (2014) recommend the inclusion of further fixed effects to control for
unobserved firm heterogeneity. In particular, they suggest including dummies for quintiles of firm
characteristics and interacting these quintile dummies with time dummies. We estimate such
a specification by including dummies for quintiles of size, book-to-market, and momentum, and
interacting each of these quintile dummies with monthly time dummies. This specification, which
includes 1,260 time/characteristic quintile interactions, continues to yield significant crisis-period
returns for high-CSR firms. The coefficient on the CSR/crisis interaction is 0.0185 (p = 0.00) for
raw returns and 0.0154 (p = 0.00) for abnormal returns (see the Internet Appendix).
23 Similarly, the trust component of the Global Competitiveness Index produced by the World
Economic Forum was still lower in September 2013 (5.54) than in September 2008 (5.65). The
abnormal returns either if prices already adjusted to the overall decline in trust
during the crisis. That is, any benefits of being trustworthy when overall trust
is low should now be reflected in the share price. Operating performance, on the
other hand, may well be affected during both the crisis and post-crisis periods.
In subsequent sections, we provide evidence that this is indeed the case.
In the specifications reported in Panel A of Table IV, we hold CSR constant
as of year-end 2006 to determine whether CSR measured before the onset of
the crisis has an effect on returns during and after the crisis. In an alterna-
tive specification, we also allow CSR to vary over time as new information on
CSR becomes available. That is, we match 2007 returns with year-end 2006
CSR, 2008 returns with year-end 2007 CSR, etc. These specifications allow
us to gauge whether updated CSR affects subsequent returns outside the cri-
sis period. These models yield similar results to those reported in Panel A of
Table IV: crisis-period returns increase with CSR, but there is no effect of CSR
on returns subsequent or prior to the crisis.
Finally, we construct a hedge portfolio that goes long in firms in the highest
quartile of CSR firms and short in firms in the lowest quartile, updating the
portfolio composition on an annual basis as new CSR information becomes
available. This portfolio earns excess returns (adjusted for four-factor loadings)
of 74 basis points per month during the crisis period, while the excess returns
are insignificant in the four years prior to and after the crisis (reported in the
Internet Appendix).
D. Excess Returns and CSR during a Shock to the Supply of Credit

We next investigate whether our results could be due to a shock to the
supply of credit, rather than a shock to market-wide trust. Starting in July
of 2007, the weakening solvency of the banking sector led to a substantial
increase in LIBOR rates, which had a strong negative impact on the ability of
firms to borrow (see, e.g., Duchin, Ozbas, and Sensoy (2010) and Ivashina and
Scharfstein (2010)). This shock to the supply of credit persisted until at least
March 2009, which is the end of the crisis period in our prior tests. If high-
CSR firms earned excess returns during the crisis because investors believed
that these firms were better able to weather the credit crunch, our test could
be picking up this effect instead of the ability to weather a shock to trust. To
investigate this possibility, we test whether CSR is related to returns in the
period July 2007 through July 2008, when the shock to credit supply already
happened but the shock to trust had not yet occurred (the Edelman Trust
Barometer shows no decline from early 2007 to early 2008; see also Sapienza
and Zingales (2012)).
For this exercise, we augment the specification of model (1) with an additional
interaction term between CSR measured at year-end 2006 and a dummy (shock
to credit) equal to one during the July 2007 to July 2008 period. The results,
Edelman Trust Barometer, in contrast, does show that trust has recovered. It was 58% in the
survey released in early 2014, which is the same as the figure released in early 2008.
which are reported in Panel B of Table IV, indicate that there is no significant
relation between CSR and either raw or abnormal returns in this earlier period.
The coefficients of the CSR/Crisis interactions remain virtually unchanged
from those in Panel A, and they are always significantly larger than those of
the CSR/credit shock interactions. Thus, a shock to credit supply is unlikely to
explain the positive association between social capital and stock returns during
the crisis documented in Table II and in Panel A of Table IV.
E. Regional Trust and the Relation between CSR and Returns

Our interpretation of the excess crisis-period returns for high-CSR firms is
that such firms build social capital through their CSR activities, which pays
off when there is a shock to overall trust. In this section, we provide evidence
for this interpretation by linking the crisis-period returns earned by high-CSR
firms to regional variation in trust across the United States.
We obtain data on regional variation in trust from the 2006 General Social
Survey (GSS) conducted by the National Opinion Research Center (NORC) at
the University of Chicago (see also Kelly (2015)). This survey asks a random
sample of Americans a large number of questions related to various aspects
of society, including: “Generally speaking, would you say that most people
can be trusted or that you can’t be too careful in dealing with people?” In
2006, the survey covered 3,929 responses to this question. After removing 192
respondents who state “Depends,” 34% of the respondents reply that people can
be trusted while the remainder reply that people cannot be trusted. There is
substantial variation in trust across the nine regions in which respondents are
classified. For example, only 26% of the respondents in the West South Central
region reply that they can trust people compared to 43% in New England and
the Mountain region.
We exploit this cross-sectional variation to explore whether regional differ-
ences in trust affect the returns earned by high-CSR firms by matching the
regional trust averages with the regions in which the firms are headquartered.
Our primary hypothesis follows the work of Putnam (2000), who argues that
an agent’s social capital is more valuable in a society where overall social cap-
ital is higher. Framing this argument in our context, in regions where people
have a lower propensity to trust, CSR activities are less likely to be viewed
by investors and other stakeholders as trust-enhancing activities; instead they
may be perceived as window dressing and less genuine activities. As such,
they are less likely to pay off. If more stakeholders are based in the region
where the firm is headquartered, we can use this regional variation in trust to
directly test whether trust matters more where it should—in more trusting re-
gions. Employees, customers, and other stakeholders in more trusting regions
are more likely to reward trustworthy firms, for example, by working harder
and maintaining strong buying relationships, leading to higher crisis-period
returns. Additionally, if investors hold local companies (see, e.g., Coval and
Moskowitz (1999)) and prices are influenced by local investors (see, e.g., Hong,
Kubik, and Stein (2008)), more trustworthy firms may also be able to raise
more capital and achieve higher valuations during the crisis.
In Panel C of Table IV, we repeat our prior analyses, but now allow the effect
of CSR on returns to vary depending on whether the firm is headquartered
in a high- or low-trust region. The results indicate that crisis-period returns
are more affected by CSR in high-trust regions compared to low-trust regions.
For raw returns, increasing CSR by one-standard-deviation is associated with
monthly excess returns of 1.05% in high-trust regions, but only 59 basis points
in low-trust regions (if we allow for a differential effect of trust on returns dur-
ing the crisis and post-crisis periods, these numbers change to 63 basis points
and 96 basis points, still a considerable difference). For abnormal returns, there
is also a substantial difference between the two sets of regions although it is of
a smaller magnitude.
Two caveats are in order. First, an individual’s ability to trust people may
be different from her ability to trust firms, and, hence, our findings should
be interpreted with this caution in mind. Second, our test assumes that the
survey response reflects an individual’s propensity to trust people (and other
agents) and, as such, her willingness to respond to a firm’s CSR efforts. This
propensity, if it is an inherent personal characteristic, should not vary (much)
over time, which is indeed what we find: it has declined very gradually from
38% in 2000 to 34% in 2014,24 but did not shift dramatically around the crisis.
We also find a similar stability in trust when splitting the sample into high-
and low-trust regions. The fact that prior work shows that regional variation in
trust impacts economic and financial development also suggests that the level
of regional trust is indeed a persistent feature (e.g., Knack and Keefer (1997)
and Guiso, Sapienza, and Zingales (2004)). The stability of this measure over
time is thus consistent with the above result that a firm’s CSR efforts pay off
less in low-trust areas. As such, it is at odds with the alternative view that a
firm’s CSR efforts are particularly valuable in areas where the propensity to
trust is low and that individuals in these regions can be persuaded to become
more trusting.
Overall, the evidence in this section indicates that the impact of CSR on
returns during the crisis period is related to the general level of trust in the
area where the company is located and is consistent with the view that the link
between returns and CSR during the crisis operates through the trust channel.
F. Elements of CSR and Returns

Next, we examine whether it is a firm’s social capital in aggregate (CSR) or
a specific component of CSR that is important for crisis-period returns. At the
outset of this paper, we argue that a firm can build social capital through a
variety of activities and that such activities can enhance the trust of all of a
firm’s stakeholders. For example, customers may reward firms for treating their
24 This gradual decline in trust is consistent with Putnam’s (2000) observation that trust has
been declining in the United States over the last several decades.
employees better, while employees may work harder because the company cares
more about its community or the environment. It is possible, however, that some
aspects of CSR are more important to building trust than others, which could
affect the strength of their relation with returns. To test this conjecture we
disaggregate CSR into two components: those that speak mainly to internal
stakeholders (Employee Relations and Diversity) and those that speak mainly
to external stakeholders (Community, Human Rights, and Environment).
Our results are reported in Panel D of Table IV using the same firm fixed
effects specification as before. Both components of CSR are significant in ex-
plaining crisis-period raw and abnormal stock returns, which indicates that
investors view a CSR focus on both internal and external stakeholders as valu-
able during the 2008–2009 financial crisis.25 In terms of economic significance,
both elements of CSR are of similar importance. An increase in Internal Stake-
holder CSR of one-standard-deviation (0.310) is associated with a 0.43 percent-
age point higher monthly abnormal return during the crisis, while an increase
in External Stakeholder CSR of one-standard-deviation (0.189) is associated
with a 0.35 percentage point higher abnormal return.
G. Further Robustness Tests

In this section, we report the results of various additional tests conducted to
determine whether our main findings are robust. We first focus on measuring
CSR performance at different points in time. In our baseline models reported
in prior tables, we measure CSR performance at the end of 2006, more than one
year before the onset of the shock to trust and several months before LIBOR
rates started rising. It is possible that some corporate managers anticipated
a potential slowdown given the heady returns for asset prices in general in
2006, and started adjusting their CSR activities in 2006 accordingly. It is also
possible that only those firms that were able to cope better with the crisis
were the ones to adjust their CSR activities upwards. While we control for
observables that could potentially affect crisis-period returns, if CSR at the
end of 2006 is correlated with some unobservable measure of the ability to
withstand a shock to trust, then the results we report may not be due to social
capital and CSR but rather to some other factor.
To address this concern, we investigate whether firm CSR scores measured
in 2005 are positively related to crisis-period stock returns since 2005 clearly
precedes any fears of a financial crisis. In the first two columns of Table V, we
re-estimate model (1) above using CSR measured in 2005 as the variable of
interest. The effect is of comparable magnitude to the effect using 2006 CSR.
We next conduct the same test using 2007 CSR data. As reported in columns
(3) and (4), our findings continue to hold.
It is also possible that high-CSR firms performed well during the crisis be-
cause prior CSR activities were actually negative NPV projects, and firms
25 Note that the scores for the Internal and External CSR categories are not highly correlated
(ρ=0.12) and thus our results to do not mechanically follow from the aggregate CSR score.
Table V
Crisis-Period Returns and CSR: Robustness
This table presents results of estimating the following panel regression model
Returni,t = bo + b1 C SRi × Crisist + b2 C SRi × Post-Crisist + b3 Xi,t−1 + Time Dummies + Firm Fixed Effects + ei,t ,
where Returni,t is the monthly raw or market-model adjusted return and CSRi is our proxy for CSR, measured at year-end 2005, 2007, or 2008.
Crisist is a dummy variable set to one in the period August 2008 to March 2009, Post-Crisist is a dummy variable set to one in the period April 2009
to December 2013, and Xi,t–1 is a vector of control variables. The control variables included in all models are the same as those employed in Table II
and they include firm financial characteristics as well as factor loadings. The financial characteristics based on accounting data are updated three
months after each fiscal year-end. The characteristics based on market data (momentum, size, market-to-book, factor loadings) are updated monthly.
Factor loadings are re-estimated each month based on the previous 60 months’ data. The firm’s CSR measure is absorbed by the firm fixed effect.
The regression is estimated over the period 2007–2013. Financial firms are removed from the sample. Micro-cap firms with a market capitalization
below $250 million (in 2007 dollars) are removed from the sample in models (1) through (6), but included in models (7) and (8). Except when otherwise
indicated, numbers in parentheses are heteroskedasticity-consistent standard errors, clustered at the firm level. ***, **, and * indicate that the
parameter estimate is significantly different from zero at the 1%, 5%, and 10% level, respectively.
CSR2005 CSR 2007 CSR2008 Including Micro-Cap Firms
Raw return Abnormal return Raw return Abnormal return Raw return Abnormal return Raw return Abnormal return
Variable (1) (2) (3) (4) (5) (6) (7) (8)
CSR × Crisis 0.0149*** 0.0097** 0.0181*** 0.0122*** 0.0180*** 0.0129*** 0.0206*** 0.0163***
(0.0043) (0.0043) (0.0038) (0.0038) (0.0036) (0.0036) (0.0038) (0.0038)
CSR × Post-Crisis 0.0022 0.0023 0.0023 0.0034 0.0030 0.0036 0.0021 0.0025
(0.0025) (0.0027) (0.0022) (0.0024) (0.0021) (0.0024) (0.0023) (0.0025)
Social Capital, Trust, and Firm Performance
CSR × (Crisis – 0.0127 0.0076 0.0158 0.0088 0.0150 0.0093 0.0185 0.0138
Post-Crisis)
p-Value (0.00) (0.05) (0.00) (0.01) (0.00) (0.00) (0.00) (0.00)
N 111,056 111,056 133,218 133,218 134,194 134,194 133,403 133,403
Adj. R2 0.29 0.07 0.28 0.06 0.28 0.06 0.27 0.07
1813
were forced to cut these activities during the crisis. If CSR is just one element
of excess investment, then the level of CSR could also proxy for the extent of
overinvestment in the firm as whole. Thus, it could be the case that firms that
engaged more in non-value-maximizing behavior pre-crisis performed better
during the crisis simply because they had more excesses that could be trimmed.
To test this conjecture, in models (5) and (6) of Table V, we examine whether
our results hold when CSR is measured at year-end 2008, when these excesses
would arguably have already been cut. Our findings persist: high CSR levels
measured in the depth of the crisis are still associated with higher crisis-period
returns.
Overall, our results are not sensitive to the time period in which CSR in-
vestments are measured. The main reason for this lack of sensitivity is that
CSR levels are relatively persistent over time. For example, the correlation in
our CSR measure between 2005 and 2006 is 0.90, the correlation between 2006
and 2007 is 0.89, and the correlation between 2005 and 2007 is 0.82.
As a second robustness test, we assess whether the decision to remove micro-
cap firms (those with equity market capitalization below $250 million) from
our sample affects our results. We excluded these firms because they typically
display very low stock market liquidity, and this factor could outweigh other
factors during the crisis. In the models presented in columns (7) and (8) of
Table V, we re-estimate our full model including these firms. Our results hold
when they are added back to our sample.
As a third robustness test, we verify that our findings are not due to the
inclusion of March 2009 as part of the crisis period. Stock markets started
recovering globally during the middle of that month and we want to ensure
that our results are not due simply to this recovery. We find that the coefficient
on CSR remains positive and significant in specifications that exclude March
2009 from the crisis return window (reported in the Internet Appendix).
IV. The Effect of CSR Investments on Operating Performance

and Capital Raising
In this section, we study the operating performance and capital-raising ac-
tivities of companies during the crisis and surrounding periods to explore in
more detail possible sources of the excess returns earned by high-CSR firms
during the crisis. We estimate a difference-in-differences model with contin-
uous treatment levels. In particular, using quarterly data, we estimate the
following regression model over the period 2007–2013 for different measures
of performance and capital raising:

Performance or Capital Measurei,t = bo + b1 C SRi,2006 × Crisist
+ b2 C SRi,2006 × Post-Crisist
+ b3 Xi,t−1 + Time Dummies
+ Firm Fixed Effects + ei,t , (2)
where CSRi,2006 is our measure of year-end 2006 CSR for firm i, Crisist is a
dummy variable set to one for the fourth quarter of 2008 and the first quarter
of 2009,26 Post-Crisist is a dummy set to one for the second quarter of 2009 until
the fourth quarter of 2013, and Xi,t–1 is a vector of control variables. All models
include quarter and firm fixed effects. Thus, if a particular firm performed well
throughout the estimation period because of some unobservable characteristics,
this effect will be captured by the fixed effect. Similarly, if the performance of
all firms varies over time (as happened during the crisis) then this will be
captured by the time dummies. To avoid problems with extreme observations,
we winsorize all performance and capital-raising variables at the 1st and 99th
percentiles. Standard errors in all models are clustered at the firm level.
Our findings are reported in Table VI. Our first performance measure is Op-
erating Return on Assets, computed as operating income divided by assets. The
interaction between CSR and the crisis-period dummy is positive and highly
significant, indicating that high-CSR firms exhibit higher profitability relative
to other companies at the end of 2008 and the beginning of 2009. In terms of
economic significance, an increase in CSR of one-standard-deviation (0.381) is
associated with an increase in profitability of 30 basis points during the cri-
sis period, which is substantial compared to average quarterly profitability of
3.1% over the estimation period and 2.2% during the crisis. Also note that the
increase in profitability for high-CSR firms persists in the post-crisis period as
well, albeit at an attenuated level. As argued previously, given that trust in cor-
porations has remained low since the end of the crisis, observing some excess
operating performance for high-CSR firms during this time is not surprising.
Next, we analyze changes in Gross Margin, defined as (sales – cost of goods
sold)/sales, to see whether high-CSR firms were able to sell their products at a
higher mark-up during the crisis. Of course, higher mark-ups could be due to
higher prices or lower costs; the gross margin just captures the net effect. As
shown in column (2) of Table VI, gross margins of high-CSR firms are higher
relative to those of low-CSR firms during the crisis. Over the crisis quarters, a
one-standard-deviation increase in CSR is associated with gross margins that
are 60 basis points higher. This effect appears small relative to average gross
margins of 40.1% during the estimation period and 38.3% during the crisis,
but the results on profitability reported in column (1) suggest that much of
this increase flows through to the bottom line. Also note that gross margins
have remained relatively higher for high-CSR firms since the end of the crisis
and, though the difference in margins is lower than during the crisis, the
change between the crisis and post-crisis periods is not statistically significant.
These findings are also consistent with the work of Albuquerque, Durnev, and
Koskinen (2015), who suggest that high-CSR firms have higher profit margins.
26 For firms whose fiscal quarters do not overlap with the normal division of a calendar year in
quarters, we consider all quarters ending in October 2008 to March 2009 as crisis quarters. We do
not include the quarter ending September 2008 as a crisis quarter because most of the performance
for that quarter precedes the Lehman bankruptcy.
Table VI
Operating Performance, Employee Growth, Capital Raising, and CSR
Surrounding the Crisis
Models (1)–(5) and (8) and (9) are regressions of various measures of performance and capital
raising using the specification
Outcomei,t = bo + b1 C SRi,2006 × Crisist + b2 C SRi,2006 × Post-Crisist + b3 Xi,t−1
+ Time Dummies + Firm Fixed Effects + ei,t ,
where CSRi,2006 is our measure of CSR computed as of year-end 2006, Crisist is a dummy variable
set equal to one for quarters ending from October 2008 to March 2009, Post-Crisist is a dummy
variable set to one for quarters ending from April 2009 to December 2013, and Xi,t–1 is a vector of
control variables. In the models of performance, we control for the log of total assets. In the models
of security issuance, we control for the log of total assets, the ratio of cash holdings to assets lagged
one quarter, the ratio of total debt to assets lagged one quarter, and the ratio of operating income to
assets. The performance measures are: Operating Return on Assets, measured as operating income
to assets, Gross Margin, measured as (sales – cost of goods sold) / sales, Sales Growth, measured
as the percentage change in sales from the previous quarter, Accounts Receivable divided by Sales,
and Sales per Employee. Measures of capital raising are: long-term debt issuance divided by assets
and equity issuance divided by assets. All data items are from quarterly Compustat, except number
of employees, which is from the annual Compustat database. Time dummies are specified at the
quarterly level. All performance and security issuance measures are winsorized at the 1st and
99th percentiles.Models (6) and (7) contain regressions of the model
Employment Growthi,t = bo + b1 C SRi,2006 × 2008/2009 + b2 C SRi,2006 × 2010/2013
+ b3 LogAssets + Time Dummies + Firm Fixed Effects + ei,t ,
where employment growth is the percentage growth in employees relative to the prior year,
2008/2009 is a dummy for observations in 2008 and 2009, and 2010/2013 is a dummy for ob-
servations in 2010–2013. Data are from the annual Compustat database and time dummies are
specified at the annual level. In model (6), employment growth is winsorized at the 1st and 99th
percentiles. In both panels, the regression is estimated over the period 2007–2013. Financial firms
and firms with a market capitalization below $250 million as of year-end 2007 are removed from
the sample. Standard errors, clustered at the firm level, are in parentheses. ***, **, and * indicate
that the parameter estimate is significantly different from zero at the 1%, 5%, and 10% level,
respectively.
Operating Gross Sales Accounts Sales per

Return on Margin Growth Receivable / Employee
Assets (in %) (%) (%) Sales (%) (in $000’s)
Variable (1) (2) (3) (4) (5)
CSR × Crisis 0.797*** 1.562*** 6.690*** −0.474 38.434***

(0.185) (0.511) (1.111) (0.962) (10.183)
CSR × Post-Crisis 0.282** 1.256*** 1.285*** 0.635 19.586**
(0.117) (0.442) (0.398) (1.815) (10.125)
p-Value (Crisis – 0.00 0.54 0.00 0.53 0.00
Post-Crisis)
Firm fixed effects Yes Yes Yes Yes Yes
Quarter fixed effects Yes Yes Yes Yes Yes
Standard errors Firm Firm Firm Firm Firm
clustered by
N 43,302 43,319 43,918 43,622 42,923
Adj. R2 0.636 0.875 0.044 0.770 0.769
(Continued)
Table VI—Continued
Employee Growth in
Employee Growth –50% and +100% range
Variable (6) (7)
CSR × 2008/2009 1.999 2.285*

(1.978) (1.253)
CSR × 2010/2013 0.031 0.792
(1.911) (1.246)
p-Value (2008/2009 – 2010/2013) 0.09 0.11
Year fixed effects Yes Yes
N 10,604 10,439
Adj. R2 0.142 0.185
Debt Equity Issuance/Assets

Issuance/Assets (%) (%)
Variable (8) (9)
CSR × Crisis 0.491** −0.058

(0.236) (0.051)
CSR × Post-Crisis 0.280 −0.034
(0.217) (0.034)
p-Value (Crisis – Post-Crisis) 0.33 0.60
Quarter fixed effects Yes Yes
Control variables Yes Yes
N 38,397 38,719
Adj. R2 0.33 0.15
One concern is that the higher mark-ups documented in column (2) may
be associated with lower sales growth. This is what we study in the model
reported in column (3), where we employ sales growth, computed as the
percentage growth in sales over the previous quarter, as the dependent vari-
able. Interestingly, high-CSR firms experience higher sales growth during the
crisis compared to other firms: a one-standard-deviation increase in CSR is
associated with 2.55 percentage points greater sales growth. This is a con-
siderable effect, given mean quarterly sales growth of 3.24% over the sample
period, and a mean decline in sales of 6.91% during the two crisis quarters.
Taken together, the findings of columns (2) and (3) indicate that, during the
crisis, high-CSR firms experienced lower declines in sales than other firms,
despite charging higher mark-ups. This suggests that the customers of these
firms were more willing to “stick” with the company during this period. Note
from column (3) that the higher level of sales growth for high-CSR firms also
persists after the crisis, although the magnitude of the effect is substantially
lower than during the crisis.
To examine the customer channel in greater depth, we study changes
in accounts receivable as a fraction of sales around the crisis. The results
reported in column (4) show no significant effect. Thus, there is no evidence

that the stronger sales growth of high-CSR firms is due to increased credit
sales. Customers of high-CSR firms are not paying their invoices any faster
during the crisis either.
In sum, the operating performance results discussed up to this point suggest
that one of the channels through which high-CSR firms earn excess returns dur-
ing the crisis period is the willingness of customers to continue supporting these
firms, as reflected in higher sales growth and an acceptance of higher mark-ups.
We now turn to the employee channel and study whether high-CSR firms
achieved higher sales per employee in the crisis period. As illustrated in column
(5) of Table VI, there is a positive association between CSR and employee pro-
ductivity during the crisis. The coefficient of 38.434 suggests that an increase
in CSR of one-standard-deviation is associated with $14,643 higher quarterly
sales per employee during the crisis. The mean (median) firm over the esti-
mation period has sales per employee of $131,484 ($75,282), with a standard
deviation of $323,585, indicating that the impact of CSR on employee produc-
tivity is considerable. This result suggests an additional channel through which
CSR affects performance. Note that this effect also persists after the crisis, but
at half the rate.27
One shortcoming of our analysis of sales per employee is that the number
of employees is only available on Compustat at an annual level. Thus, we
divide quarterly sales by the number of employees at year-end to compute
sales per employee. For example, sales per employee for the first quarter of
2009 are computed using the number of employees reported for year-end 2009.
We also verify that our findings remain unchanged when we lag the number
of employees by one year. Finally, to verify that higher sales per employee are
not due to employee layoffs, we compute growth in the number of employees on
an annual basis and estimate models of employee growth as a function of CSR,
year dummies, firm fixed effects, the interaction between CSR and a dummy for
2008/2009, and the interaction between CSR and a dummy for 2010–2013. This
regression is similar to those estimated for the performance measures, except
that it is estimated using annual instead of quarterly data. As illustrated in
column (6) of Table VI, there is no evidence of higher employee layoffs for
high-CSR firms in 2008 or 2009. In fact, if we remove firms that double their
employees or lose half their employees in a year, we find some evidence that
high-CSR firms experience more employee growth in 2008 and 2009 relative to
low-CSR firms, as illustrated in column (7).
27 We also estimated these models without the time dummies but with dummies for the crisis
and post-crisis periods. The coefficients on these dummies allow us to determine the effect of the
crisis on performance for firms with a CSR score of zero. In general, we find evidence of a strong
decline in performance during the crisis and of a substantial recovery in the post-crisis period.
For example, during the crisis quarters, firms with a CSR score of zero experienced a decline in
operating return to assets of 0.91 percentage points, a decline in gross margin of 1.4 percentage
points, and a decline in sales growth of 9.8 percentage points. In the post-crisis period, profitability
improved by 0.76 percentage points, gross margins increased by 0.70 percentage points, and sales
grew by 9.3 percentage points relative to the crisis period.
Next, we focus on the investor channel and study capital raising during and
surrounding the crisis. We divide both long-term debt and equity issues by
total assets and relate these debt or equity issue measures to CSR activities as
in equation (2) above. We report results of these specifications in columns (8)
and (9) of Table VI. As illustrated in column (8), high-CSR firms raised more
debt during the crisis, albeit the economic effect is modest. Increasing CSR
by one-standard-deviation increases debt issuances relative to assets by 0.19
percentage points while average debt issuance is 2.61% over the sample period
and 2.26% during the crisis. Also note that the effect of CSR on debt capital
raising becomes insignificant in the post-crisis period. Equity issuances, which
are studied in column (9), are not related to CSR around the crisis period.
In our final set of tests, we relate the measures of operating performance
presented in Table VI to the stock price performance documented previously.
To do so, we re-estimate our regression model of returns in Table II, but add the
actual performance measures achieved during the crisis period as explanatory
variables. The goal of this exercise is not to predict returns but rather to assess
the extent to which the cross-sectional variation in crisis-period returns can
be explained by concurrent operating performance. Actual performance is com-
puted as average performance over the quarters ending from October 2008 to
March 2009. We include profitability, gross margin, sales growth, and sales per
employee as performance measures. These models (see the Internet Appendix)
show that profitability, gross margin, and sales growth all affect crisis-period
(abnormal) returns, with the economic effect of profitability being the largest.
Increasing profitability by one-standard-deviation is associated with incremen-
tal crisis-period returns of 5.88 percentage points; the economic effect of gross
margin is somewhat smaller while the effect of sales growth is around half of
that. In these models, the coefficient on CSR is reduced by about half for raw
returns and one quarter for abnormal returns. For raw returns, the coefficient
on CSR is no longer significant, but it is significant at the 5% level for abnor-
mal returns. This suggests that part of the excess returns earned by high-CSR
firms during the crisis are due to their superior operating performance (related
to stakeholder trust). The unexplained variation in returns could be due to
the direct effect of social capital on shareholder trust (see Guiso, Sapienza, and
Zingales (2004, 2008)) although we recognize that other performance measures
may also matter for stock returns, which would attenuate this effect.
Overall, the evidence reported in this section broadly suggests that some
of the increased returns to high-CSR firms accrue through the customer and
employee channels. There is modest evidence in support of increased debt
capital raising. It is worth noting that, even after controlling for these real
effects, CSR continues to have a persistent, albeit reduced, impact on crisis-
period returns.
V. Conclusion
This paper provides evidence that firm-specific social capital, built up
through CSR activities, pays off during a period when the importance of trust
increases unexpectedly, namely the 2008–2009 financial crisis. In particular,

we find that firms with high CSR ratings outperform firms with low CSR rat-
ings during the crisis by at least four percentage points, after controlling for a
variety of firm characteristics and risk factors. We also find that the excess re-
turns are higher for firms headquartered in regions where individuals are more
trusting. There is no difference in stock return performance between high- and
low-CSR firms during the recovery period after the crisis. Collectively, these
results suggest that increased social capital resulting from CSR activities mat-
ters predominantly in periods when trust in corporations at large has eroded,
and that during normal times any benefits of social capital are already imbed-
ded in a firm’s share price. The lack of a reversal in returns in the post-crisis
period suggests that being trustworthy has remained important, which is in
line with survey data that report continuing low levels of trust in corporations
and the stock market.
We also examine the mechanisms through which higher CSR levels might
generate excess returns during the crisis and find that high-CSR firms benefit
through higher profitability, margins, sales growth, and employee productivity
relative to low-CSR firms. Some of these effects also persist in the post-crisis
period, but at lower levels of economic and statistical significance, again con-
sistent with survey evidence suggesting that trust remains relatively low.
Overall, our results suggest that the building of firm-specific social capital
can be thought of as an insurance policy that pays off when investors and the
overall economy face a severe crisis of confidence. Our work also indicates that
social capital, in addition to financial capital, can be an important determinant
of firm performance, and identifies circumstances under which CSR can be
beneficial for firm value.
Two caveats are in order. First, as with most empirical work, unobserved
time-varying firm heterogeneity could explain our findings, but the fact that
our results survive the inclusion of a large battery of control variables as well as
firm and time fixed effects, and hold during another period when trust suffered
a shock, mitigates this concern. Second, in constructing our proxy for social
capital at the firm level, we rely on prior literature suggesting a link between
CSR and the formation of social capital. However, there may be other channels
through which firms can build social capital and increase trust. Examining
these channels and studying the relative efficacy of the CSR channel compared
to other channels would be a fruitful avenue for further work.
Initial submission: January 26, 2015; Accepted: September 25, 2016

Editors: Bruno Biais, Michael R. Roberts, and Kenneth J. Singleton
REFERENCES
Albuquerque, Rui, Art Durnev, and Yrjo Koskinen, 2015, Corporate social responsibility and firm
risk: Theory and empirical evidence, Working paper, Boston University, University of Iowa,
and University of Calgary.
Almeida, Heitor, Murillo Campello, Bruno Laranjeira, and Scott Weisbenner, 2012, Corporate debt
maturity and the real effects of the 2007 credit crisis, Critical Finance Review 1, 3–58.
Arrow, Kenneth J., 1972, Gifts and exchanges, Philosophy and Public Affairs 1, 343–362.
Bebchuk, Lucian, Alma Cohen, and Allen Ferrell, 2009, What matters in corporate governance?
Review of Financial Studies 22, 783–827.
Bénabou, Roland, and Jean Tirole, 2010, Individual and corporate social responsibility, Economica
77, 1–19.
Borisov, Alexander, Eitan Goldman, and Nandini Gupta, 2016, The corporate value of (corrupt)
lobbying, Review of Financial Studies 29, 1039–1071.
Carlin, Bruce I., Florin Dorobantu, and S. Viswanathan, 2009, Public trust, the law, and financial
investment, Journal of Financial Economics 92, 321–341.
Cheng, Ing-Haw, Harrison Hong, and Kelly Shue, 2016, Do managers do good with other peoples’
money? Working paper, Dartmouth College, Princeton University, and University of Chicago.
Coleman, James S., 1988, Social capital in the creation of human capital, American Journal of
Sociology 94 (Supplement), S95–S120.
Coleman, James S., 1990, Foundations of Social Theory (Harvard University Press, Cambridge,
MA).
Coval, Joshua D., and Tobias J. Moskowitz, 1999, Home bias at home: Local equity preference in
domestic portfolios, Journal of Finance 54, 2045–2073.
Daniel, Kent, and Sheridan Titman, 1997, Evidence on the characteristics of cross-sectional vari-
ation in stock returns, Journal of Finance 52, 1–33.
Dasgupta, Partha, 1988, Trust as a commodity, in Diego Gambetta, ed.: Trust: Making and Break-
ing Cooperative Relations (Basil Blackwell, Oxford, UK).
Deng, Xin, Jun-Koo Kang, and Buen Sin Low, 2013, Corporate social responsibility and stakeholder
value maximization: Evidence from mergers, Journal of Financial Economics 110, 87–109.
Di Giuli, Alberta, and Leonard Kostovetsky, 2014, Are red or blue companies more likely to go
green? Politics and corporate social responsibility, Journal of Financial Economics 111, 158–
180.
Duchin, Ran, Oguzhan Ozbas, and Berk A. Sensoy, 2010, Costly external finance, corporate invest-
ment, and the subprime mortgage credit crisis, Journal of Financial Economics 97, 418–435.
Eccles, Robert G., Ioannis Ioannou, and George Serafeim, 2014, The impact of corporate sustain-
ability on organizational processes and performance, Management Science 60, 2835–2857.
Edmans, Alex, 2011, Does the stock market fully value intangibles? Employee satisfaction and
equity prices, Journal of Financial Economics 101, 621–640.
Erhard, Werner, and Michael C. Jensen, 2015, Putting integrity into finance: A purely positive
approach, Working paper, Harvard Business School.
Erhard, Werner, Michael C. Jensen, and Steve Zaffron, 2009, Integrity: A positive model that in-
corporates the normative phenomena of morality, ethics, and legality, Working paper, Harvard
Business School.
Fama, Eugene F., and Kenneth R. French, 1992, The cross-section of expected stock returns,
Journal of Finance 47, 427–465.
Ferrell, Allen, Hao Liang, and Luc Renneboog, 2016, Socially responsible firms, Journal of Finan-
cial Economics 122, 585–606.
Fitzgerald, Niall, 2003, CSR: Rebuilding Trust in Business. A Perspective on Corporate Social
Responsibility in the 21st Century, speech by Unilever’s Chairman, Unilever and London
Business School, London, UK.
Flammer, Caroline, 2015, Does corporate social responsibility lead to superior financial perfor-
mance? A regression discontinuity approach, Management Science 61, 2549–2568.
Fukuyama, Francis, 1995, Trust (Free Press, New York, NY).
Gambetta, Diego, 1988, Can we trust trust? in Diego Gambetta, ed.: Trust: Making and Breaking
Cooperative Relations (Basil Blackwell, Oxford, UK).
Gao, Feng, Ling Lei Lisic, and Ivy Zhang, 2014, Commitment to social good and insider trading,
Journal of Accounting and Economics 57, 149–175.
Glaeser, Edward L., David Laibson, and Bruce Sacerdote, 2002, An economic approach to social
capital, Economic Journal 112, F437–F458.
Godfrey, Paul C., Craig B. Merrill, and Jared M. Hansen, 2009, The relationship between corpo-
rate social responsibility and shareholder value: An empirical test of the risk management
hypothesis, Strategic Management Journal 30, 425–445.
Gormley, Todd A., and David A. Matsa, 2014, Common errors: How to (and not to) control for
unobserved heterogeneity, Review of Financial Studies 27, 617–661.
Gouldner, Alvin W., 1960, The norm or reciprocity: A preliminary statement, American Sociological
Review 25, 161–178.
Goyal, Amit, and Pedro Santa-Clara, 2003, Idiosyncratic risk matters! Journal of Finance 58,
975–1008.
Guiso, Luigo, Paola Sapienza, and Luigi Zingales, 2004, The role of social capital in financial
development, American Economic Review 94, 526–556.
Guiso, Luigo, Paola Sapienza, and Luigi Zingales, 2008, Trusting the stock market, Journal of
Finance 63, 2557–2600.
Guiso, Luigo, Paola Sapienza, and Luigi Zingales, 2011, Civic capital as the missing link, in Jess
Benhabib, Alberto Bisin, and Matthew O. Jackson, eds.: Social Economics Handbook (North
Holland, Amsterdam).
Guiso, Luigo, Paola Sapienza, and Luigi Zingales, 2015, The value of corporate culture, Journal of
Financial Economics 117, 60–76.
Gurria, Angel, 2009, Keynote Remarks by OECD Secretary-General, Available at http://www.oecd.
org/newsroom/respondingtotheglobaleconomiccrisisoecdsroleinpromotingopenmarketsandjob
creation.htm. Accessed April 19, 2017.
Harford, Jarrad, Sandy Klasa, and William Maxwell, 2014, Refinancing risk and cash holdings,
Journal of Finance 69, 975–1012.
Hong, Harrison G., and Leonard Kostovetsky, 2012, Red and blue investing: Values and finance,
Journal of Financial Economics 103, 1–19.
Hong, Harrison G., Jeffrey D. Kubik, and Jeremy C. Stein, 2008, The only game in town: Stock-price
consequences of local bias, Journal of Financial Economics 90, 20–37.
Hong, Harrison G., and Inessa Liskovich, 2016, Crime, punishment and the value of corporate
social responsibility, Working paper, Princeton University.
Ivashina, Victoria, and David Scharfstein, 2010, Bank lending during the financial crisis of 2008,
Journal of Financial Economics 97, 319–338.
Kelly, Peter, 2015, Dividends and trust, Working paper, University of Notre Dame.
Kim, Yongtae, Myung Seok Park, and Benson Wier, 2012, Is earnings quality associated with
corporate social responsibility? The Accounting Review 87, 761–796.
Kitzmueller, Markus, and Jay Shimshack, 2012, Economic perspectives on corporate social respon-
sibility, Journal of Economic Literature 50, 51–84.
Knack, Stephen, and Philip Keefer, 1997, Does social capital have an economic payoff? A cross-
country investigation, Quarterly Journal of Economics 112, 1251–1288.
Krüger, Philipp, 2015, Corporate goodness and shareholder wealth, Journal of Financial Eco-
nomics 115, 304–329.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert W. Vishny, 1997, Trust
in large organizations, American Economic Review Papers and Proceedings 87, 333–338.
Leana, Carrie R., and Harry J. Van Buren, III, 1999, Organizational social capital and employment
practices, Academy of Management Review 24, 538–555.
Lin, Nan, 2001, Social Capital: A Theory of Social Structure and Action (Cambridge University
Press, Cambridge, UK).
Lins, Karl V., Paolo Volpin, and Hannes Wagner, 2013, Does family control matter? International
evidence from the 2088–2009 financial crisis, Review of Financial Studies 26, 2583–2619.
Margolis, Joshua D., Hillary Anger Elfenbein, and James P. Walsh, 2009, Does it pay to be good . . .
and does it matter? A meta-analysis of the relationship between corporate social and financial
performance, Working paper, Harvard University, Washington University in St. Louis, and
University of Michigan.
Minor, Dylan B., 2015, The value of corporate citizenship: Protection, Working paper, Harvard
Business School.
Nguyen, Tu, H.G. (Lily) Nguyen, and Xiangkang Yin, 2015, Corporate governance and corporate
financing and investment during the 2007–2008 financial crisis, Financial Management 44,
115–146.
Porter, Michael E., and Mark R. Kramer, 2006, Strategy and society: The link between competitive
advantage and corporate social responsibility, Harvard Business Review 84, 78–92.
Porter, Michael E., and Mark R. Kramer, 2011, Creating shared value, Harvard Business Review
89, 1–17.
PricewaterhouseCoopers, 2013, 16th CEO Survey, Available at http://www.pwc.com/gx/
en/ceo-survey/2013/assets/pwc-16th-global-ceo-survey jan-2013.pdf. Accessed April 17, 2017.
PricewaterhouseCoopers, 2014, 17th CEO Survey, Available at http://www.pwc.com/gx/
en/ceo-survey/2014/assets/pwc-17th-annual-global-ceo-survey-jan-2014.pdf. Accessed April
17, 2017.
Putnam, Robert D., 1993, Making Democracy Work: Civic Traditions in Modern Italy (Princeton
University Press, Princeton, NJ).
Putnam, Robert D., 2000, Bowling Alone: The Collapse and Revival of American Community (Simon
and Schuster, New York, NY).
Reich, Robert, Government needs to rebuild trust in the markets, U.S. News and World Re-
port, September 16, 2008. Available at http://www.usnews.com/opinion/articles/2008/09/16/
robert-reich-government-needs-to-rebuild-trust-in-the-markets. Accessed April 17, 2017.
Sacconi, Lorenzo, and Giacomo Degli Antoni, 2011, Social Capital, Corporate Responsibility, Eco-
nomic Behaviour and Performance (Palgrave MacMillan, New York, NY).
Sapienza, Paola, Anna Toldra-Simats, and Luigi Zingales, 2013, Understanding trust, Economic
Journal 123, 1313–1332.
Sapienza, Paola, and Luigi Zingales, 2012, A trust crisis, International Review of Finance 12,
123–131.
Scrivens, Katherine, and Conal Smith, 2013, Four interpretations of social capital: An agenda for
measurement, OECD Statistics Working paper.
Servaes, Henri, and Ane Tamayo, 2013, The impact of corporate social responsibility on the value
of the firm: The role of customer awareness, Management Science 59, 1045–1061.
Stiglitz, Joseph E., The fruit of hypocrisy, Economics Opinion, The Guardian, September 16, 2008.
Available at https://www.theguardian.com/commentisfree/2008/sep/16/economics.wallstreet.
Accessed April 17, 2017.
Solow, Robert M., 1995, Trust: The social virtues and the creation of prosperity (Book review), The
New Republic 213, 36–40.
Tonkiss, Fran, 2009, Trust, confidence and economic crisis, Intereconomics 44, 196–202.
World Business Council for Sustainable Development, 2000, Corporate Social Responsibility: Mak-
ing Good Business Sense (Geneva, Switzerland).
Supporting Information
Additional Supporting Information may be found in the online version of this
article at the publisher’s website:
Appendix S1: Internet Appendix.

Social Capital, Trust, and Firm Performance: The Value of Corporate Social Responsibility During The Financial Crisis

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Social Capital, Trust, and Firm Performance: The Value of Corporate Social Responsibility During The Financial Crisis

Uploaded by

Copyright:

Available Formats

THE JOURNAL OF FINANCE • VOL. LXXII, NO.

Social Capital, Trust, and Firm Performance:

“The present financial crisis springs from a catastrophic collapse in confi-

“Something important was destroyed in the last few months of 2008. It is

trustworthy is more valuable, such as in an unexpectedly low-trust period.

I. Trust, Social Capital, and Corporate Social Responsibility

B. Social Capital and Corporate Social Responsibility

C. Social Capital and Firm Valuation

stakeholders are more likely to “do whatever it takes” to help high-social-capital

II. Sample and Summary Statistics

by Boston-based KLD Research and Analytics, Inc. (KLD).

penalize participation in the six industries that are considered controversial,

Panel A: Summary Statistics

Mean SD (Std Dev) 25th perc. Median 75th perc.

CSR −0.165 0.381 −0.343 −0.200 0.006

Short-Term Debt 0.029 0.055 0 0.0055 0.031

Panel B: Correlation Matrix

Crisis Raw Return 0.11

−0.09 −0.10 −0.02 −0.25 −0.11 −0.21 −0.19

III. Crisis-Period Returns

Panel A: Net CSR Score: Raw and Abnormal Returns

Raw return Abnormal return Raw return Abnormal return

CSR 0.059*** 0.109*** 0.048*** 0.087***

Panel A: Net CSR Score: Raw and Abnormal Returns

Raw return Abnormal return Raw return Abnormal return

Idiosyncratic Risk −3.155*** −8.870***

Raw return Abnormal return Raw return Abnormal return

CSR2 0.0303 0.0560 0.0296 0.0480

Panel C: Controlling for Corporate Governance

Raw return Abnormal return Raw return Abnormal return

CSR 0.048*** 0.090*** 0.049*** 0.088***

B. Excess Returns and CSR during the Enron/Worldcom Fraud Scandals

Raw Return Abnormal Return Raw Return Abnormal Return

Constant 1.058*** 0.977*** 0.116 0.244

indicator variable for high-CSR firms, which is employed as an explanatory

C. Comparing Returns Inside and Outside of the Crisis Period

19 We are grateful to an anonymous referee for suggesting this test.

Returni,t = bo + b1 C SRi,2006 × Crisist + b2 C SRi,2006 × Post-Crisist + b3 Xi,t−1

Panel A: Overall CSR

Variable Raw return Abnormal return

CSR × Crisis 0.0201*** 0.0153***

Variable Raw return Abnormal return

CSR × Shock to Credit 0.0021 0.0001

Variable Raw return Abnormal return

CSR × Post-Crisis 0.0033 0.0021

Panel C: Split by High- and Low-Trust Regions

Variable Raw return Abnormal return

CSR × Crisis × High Trust 0.0275*** 0.0171**

Panel D: Internal and External Stakeholder CSR

Variable Raw return Abnormal return

Int. Stakeholder CSR × Crisis 0.0225*** 0.0140***

Panel D: Internal and External Stakeholder CSR

Variable Raw return Abnormal return

Ext. Stakeholder CSR × Post-Crisis 0.0057* −0.0048

a one-standard-deviation increase in 2006 CSR (0.381) is associated with a

D. Excess Returns and CSR during a Shock to the Supply of Credit

E. Regional Trust and the Relation between CSR and Returns

F. Elements of CSR and Returns

G. Further Robustness Tests

CSR2005 CSR 2007 CSR2008 Including Micro-Cap Firms

IV. The Effect of CSR Investments on Operating Performance

Outcomei,t = bo + b1 C SRi,2006 × Crisist + b2 C SRi,2006 × Post-Crisist + b3 Xi,t−1

CSR 0.059* 0.109* 0.048* 0.087*

Idiosyncratic Risk −3.155* −8.870*

CSR 0.048* 0.090* 0.049* 0.088*

Constant 1.058* 0.977* 0.116 0.244

CSR × Crisis 0.0201* 0.0153*

CSR × Crisis × High Trust 0.0275* 0.0171

Int. Stakeholder CSR × Crisis 0.0225* 0.0140*

CSR × Crisis 0.797* 1.562* 6.690* −0.474 38.434*